WO2023195831A1 - Apparatus for determining and treating pressure sore conditions in hospital and operating method of apparatus - Google Patents

Abstract

The present disclosure relates to an apparatus for managing pressure sores and an operating method of the apparatus for managing pressure sores. The operating method comprises the steps of: capturing, by a pressure sore management apparatus, a first image of a part of a patient where a pressure sore is suspected; after a predetermined time has elapsed since the first image was captured, capturing, by the pressure sore management apparatus, a second image of a part of the patient where a pressure sore is suspected; after a predetermined time has elapsed since the second image was captured, capturing, by the pressure sore management apparatus, a third image of the opposite side of the part of the patient; obtaining an average pixel value by averaging values of a plurality of pixels included in the third image; extracting an average red (R) component value, an average green (G) component value, and an average blue (B) component value included in the average pixel value; and determining an average GB component value on the basis of the average G component value and the average B component value.

Description

Device and method of operation of the device for determining and treating pressure ulcer status in a hospital

The present disclosure relates to a device for recognizing early pressure ulcers and a method of operating the device for managing pressure ulcers.

A bedsore refers to a skin injury that occurs when the patient's back, waist, shoulders, elbows, anus around the buttocks, etc. come in contact with the floor due to the skin pressed by the patient's body weight and the blood not flowing smoothly. Generally, bedsores occur in patients who are hospitalized for a long period of time or in patients who cannot freely change positions due to cranial spinal nerve disorders. Additionally, bedsores can also occur when pressure is applied to a splint or cast. The main causes of bedsores are when the skin is continuously pressed against a hard material for a long period of time, moisture is generated in the area where the skin is pressed, lack of ventilation, lack of nutrition, or poor cleanliness or health.

When a bedsore occurs, blood circulation in the compressed area worsens, so at first the area turns pale and the surrounding area becomes red and tender. As the condition worsens, blisters form, and if it progresses further, it becomes dark and an ulcer forms in the area where gangrene occurred. It forms and produces foul-smelling secretions, reaching dangerous levels.

Since bedsores are a chronic disease and need to be managed for a long period of time, home care and outpatient treatment are often combined rather than inpatient treatment. However, due to the nature of the disease, there is a problem in that most patients have difficulty moving for outpatient treatment.

Accordingly, there is an urgent need for the development of technology that can relieve patients' discomfort and reduce economic and social costs by allowing patients with bedsores to manage and treat their bedsores without outpatient or inpatient treatment.

Meanwhile, the proportion of artificial intelligence technology being used in the medical imaging technology field is increasing significantly. In modern medicine, medical imaging is a very essential tool for effective disease diagnosis and patient treatment. With the development of imaging technology, it has become possible to acquire more sophisticated medical imaging data, and the amount of data is becoming increasingly vast. We are approaching a time when artificial intelligence technology that can predict diseases as well as treat them after the onset of diseases by systematically analyzing vast amounts of collected medical data is absolutely necessary.

The method of operating the pressure ulcer management device of the present disclosure includes the steps of: the pressure ulcer management device capturing a first image of a region suspected of having a pressure ulcer in a patient; and, after a predetermined time after capturing the first image, the pressure ulcer management device detects the pressure ulcer in the patient. Taking a second image of the suspected area, a predetermined time after taking the second image, having the pressure ulcer management device take a third image of the opposite side of the patient's area, ascites included in the third image. Obtaining an average pixel value by averaging pixel values, extracting the average R (red) component value, average G (green) component value, and average B (blue) component value included in the average pixel value, Determining an average GB component value based on the average G component value and the average B component value, and determining the first GB component value based on the first G component value and the first B component value of the pixel included in the first image. Obtaining a first subtraction value by subtracting (first GB component value/average GB component value) * average R component value from the first R component value of the pixel included in the first image, first image Obtaining a first number, which is the number of pixels among a plurality of pixels included in a plurality of pixels whose first subtraction value is greater than or equal to a predetermined threshold value, a second G component value and a second B component value of the pixels included in the second image. A step of determining a second GB component value based on, subtracting (second GB component value/average GB component value) * average R component value from the second R component value of a pixel included in the second image, thereby performing a second subtraction. Obtaining a value, obtaining a second number, which is the number of pixels among a plurality of pixels included in the second image whose second subtraction value is greater than or equal to a predetermined threshold value, and subtracting the second number from the first number. If the subtracted value is less than or equal to a predetermined threshold number, generating a signal indicating that there is a possibility of pressure ulcers in the patient.

The step of determining the average GB component value of the operating method of the pressure ulcer management device of the present disclosure includes determining the average GB component value by averaging the average G component value and the average B component value, and determining the first GB component value. The step includes determining the first GB component value by averaging the first G component value and the first B component value for each pixel included in the first image, and the step of determining the second GB component value is, It includes determining a second GB component value by averaging the second G component value and the second B component value for each pixel included in the second image.

The step of determining the first GB component value of the operating method of the pressure ulcer management device of the present disclosure includes obtaining a first average G component value that is the average of the G components of a plurality of pixels included in the first image, Obtaining a first average B component value that is the average of the B components of the plurality of pixels included, and determining a first GB component value by averaging the first average G component value and the first average B component value. And the step of determining the second GB component value includes obtaining a second average G component value that is the average of the G components of a plurality of pixels included in the second image, and the B component of the plurality of pixels included in the second image. It includes obtaining a second average B component value that is the average of the values, and determining a second GB component value by averaging the second average G component value and the second average B component value.

The step of generating a signal indicating the possibility of a pressure ulcer in a patient of the operating method of the pressure ulcer management device of the present disclosure includes subtracting the average GB component value from the first GB component value and obtaining a first difference value obtained by taking the absolute value. , obtaining a second difference value obtained by subtracting the average GB component value from the second GB component value and taking the absolute value, and the value obtained by subtracting the second number from the first number is less than or equal to a predetermined threshold number and is the first difference value. Or, when the second difference value is less than or equal to a predetermined threshold difference value, generating a signal indicating that there is a possibility of pressure ulcers in the patient.

In the method of operating the pressure ulcer management device of the present disclosure, the step of generating a signal indicating the possibility of a pressure ulcer in a patient includes: the first number is 1/2 or more of the total number of pixels in the first image, and the first number is 1/2 or more of the total number of pixels in the first image, and the second number is If the subtracted number is less than or equal to a predetermined threshold number, generating a signal indicating that there is a possibility of pressure ulcers in the patient.

Additionally, a program for implementing the method of operating the bedsore management device as described above may be recorded on a computer-readable recording medium.

The apparatus for managing bedsores and the operating method of the apparatus for managing bedsores of the present disclosure enable machine learning of the relationship between pictures of bedsores and information about bedsores to evaluate the stage of bedsores and guide dressing treatments accordingly, thereby reducing medical expenses. It is expected to reduce and increase the efficiency of medical services.

In addition, due to the recent shortage of medical personnel, there are cases where non-professional personnel must simply determine the patient's condition. However, in the case of early lesions, the signs are not clear, so it is sometimes difficult for non-professional personnel to judge the patient's condition. In addition, there are cases where the patient's condition worsens because the lesion is not detected in the early stage and appropriate treatment is not received. The pressure ulcer management device of the present disclosure can detect early lesions and enable patients to receive appropriate treatment before their condition worsens.

Figure 1 is a block diagram of a bedsore management device 100 according to an embodiment of the present disclosure.

Figure 2 is a diagram showing a bedsore management system according to an embodiment of the present disclosure.

Figure 3 is a flowchart showing the operation of a pressure ulcer management device according to an embodiment of the present disclosure.

Figure 4 is a block diagram showing a pressure ulcer management device according to an embodiment of the present disclosure.

Figure 5 is a diagram illustrating an example of an interface and process for annotating image data of existing bedsores according to an embodiment of the present disclosure.

Figure 6 is a diagram illustrating an example of an interface and process for machine learning of image data of a plurality of existing pressure ulcers according to an embodiment of the present disclosure.

Figure 7 is a block diagram showing a pressure ulcer management device according to an embodiment of the present disclosure.

Figure 8 is a flowchart showing a dressing recommendation algorithm according to an embodiment of the present disclosure.

Figure 9 is a flowchart showing a dressing recommendation algorithm according to an embodiment of the present disclosure.

Figure 10 is a flowchart showing a dressing recommendation algorithm according to an embodiment of the present disclosure.

Figure 11 shows an example of the operation flow of a bedsore management device according to an embodiment of the present disclosure.

Figure 12 shows an example of an evaluation process of a machine learning model according to an embodiment of the present disclosure.

Figure 13 is a flowchart for explaining a method of operating a bedsore management device according to an embodiment of the present disclosure.

Figure 14 is a flowchart showing the operation of a pressure ulcer management device according to an embodiment of the present disclosure.

Figure 15 is a diagram for explaining the operation of a pressure ulcer management device according to an embodiment of the present disclosure.

Figure 16 is a flowchart showing the operation of a pressure ulcer management device according to an embodiment of the present disclosure.

Advantages and features of the disclosed embodiments and methods for achieving them will become clear by referring to the embodiments described below in conjunction with the accompanying drawings. However, the present disclosure is not limited to the embodiments disclosed below and may be implemented in various different forms. The present embodiments are merely provided to ensure that the present disclosure is complete and to those skilled in the art to which the present disclosure pertains. It is only provided to fully inform the user of the scope of the invention.

Terms used in this specification will be briefly described, and the disclosed embodiments will be described in detail.

The terms used in this specification are general terms that are currently widely used as much as possible while considering the function in the present disclosure, but this may vary depending on the intention or precedent of a technician working in the related field, the emergence of new technology, etc. In addition, in certain cases, there are terms arbitrarily selected by the applicant, and in this case, the meaning will be described in detail in the description of the relevant invention. Therefore, the terms used in this disclosure should be defined based on the meaning of the term and the overall content of this disclosure, rather than simply the name of the term.

In this disclosure, singular expressions include plural expressions, unless the context clearly specifies the singular. Additionally, plural expressions include singular expressions, unless the context clearly specifies plural expressions.

When it is said that a part "includes" a certain element throughout the specification, this means that, unless specifically stated to the contrary, it does not exclude other elements but may further include other elements.

Additionally, the term “unit” used in the specification refers to a software or hardware component, and the “unit” performs certain roles. However, “wealth” is not limited to software or hardware. The “copy” may be configured to reside on an addressable storage medium and may be configured to run on one or more processors. Thus, as an example, “part” refers to software components, such as object-oriented software components, class components, and task components, processes, functions, properties, procedures, Includes subroutines, segments of program code, drivers, firmware, microcode, circuits, data, databases, data structures, tables, arrays, and variables. The functionality provided within the components and “parts” may be combined into smaller numbers of components and “parts” or may be further separated into additional components and “parts”.

According to one embodiment of the present disclosure, “unit” may be implemented with a processor and memory. The term “processor” should be interpreted broadly to include general purpose processors, central processing units (CPUs), microprocessors, digital signal processors (DSPs), controllers, microcontrollers, state machines, etc. In some contexts, “processor” may refer to an application-specific integrated circuit (ASIC), programmable logic device (PLD), field programmable gate array (FPGA), etc. The term “processor” refers to a combination of processing devices, for example, a combination of a DSP and a microprocessor, a combination of a plurality of microprocessors, a combination of one or more microprocessors in combination with a DSP core, or any other such combination of configurations. It may also refer to

The term “memory” should be interpreted broadly to include any electronic component capable of storing electronic information. The terms memory include random access memory (RAM), read-only memory (ROM), non-volatile random access memory (NVRAM), programmable read-only memory (PROM), erasable-programmable read-only memory (EPROM), electrical may refer to various types of processor-readable media, such as erasable PROM (EEPROM), flash memory, magnetic or optical data storage, registers, etc. A memory is said to be in electronic communication with a processor if the processor can read information from and/or write information to the memory. The memory integrated into the processor is in electronic communication with the processor.

Below, with reference to the attached drawings, embodiments will be described in detail so that those skilled in the art can easily implement them. In order to clearly explain the present disclosure in the drawings, parts unrelated to the description are omitted.

Figure 1 is a block diagram of a bedsore management device 100 according to an embodiment of the present disclosure.

Referring to FIG. 1 , the pressure ulcer management device 100 according to one embodiment may include a data learning unit 110 and a data recognition unit 120. Additionally, the bedsore management device 100 may include a processor and memory.

The data learning unit 110 may receive the first data and the second data and obtain a machine learning model for the relationship between the first data and the second data. The machine learning model acquired by the data learning unit 110 may be a model for generating second data using first data. For example, the data learning unit 110 may learn the relationship between image data of a plurality of existing bedsores and information about the existing bedsores. When information about bedsores is given along with image data of a plurality of bedsores, the data learning unit 110 can create a machine learning model for predicting treatment information for bedsores. In the present disclosure, image data of bedsores may be a 2D image or a 3D image.

The data recognition unit 120 may apply the image data of the current patient's bedsores to a machine learning model to output pressure ulcer-related information and/or pressure ulcer treatment information. The machine learning model can include information about criteria for what treatment should be performed according to the image of the bedsore. Additionally, the data recognition unit 120 can use image data of bedsores and results output by the machine learning model to update the machine learning model.

At least one of the data learning unit 110 and the data recognition unit 120 may be manufactured in the form of at least one hardware chip and mounted on an electronic device. For example, at least one of the data learning unit 110 and the data recognition unit 120 may be manufactured in the form of a dedicated hardware chip for artificial intelligence (AI), or an existing general-purpose processor (e.g., CPU Alternatively, it may be manufactured as part of an application processor) or a dedicated graphics processor (e.g., GPU) and may be mounted on various electronic devices as previously described.

Additionally, the data learning unit 110 and the data recognition unit 120 may be mounted on separate electronic devices. For example, one of the data learning unit 110 and the data recognition unit 120 may be included in the electronic device, and the other may be included in the server. In addition, the data learning unit 110 and the data recognition unit 120 may provide machine learning model information built by the data learning unit 110 to the data recognition unit 120 through wired or wireless communication, and may recognize the data. Data input to the unit 120 may be provided to the data learning unit 110 as additional learning data.

Meanwhile, at least one of the data learning unit 110 and the data recognition unit 120 may be implemented as a software module. When at least one of the data learning unit 110 and the data recognition unit 120 is implemented as a software module (or a program module including instructions), the software module is a memory or a readable non-readable module that can be read by a computer. It may be stored on a temporary readable recording medium (non-transitory computer readable media). Additionally, in this case, at least one software module may be provided by an operating system (OS) or a predetermined application. Alternatively, part of at least one software module may be provided by an operating system (OS), and the remaining part may be provided by a predetermined application.

The data learning unit 110 according to an embodiment of the present disclosure may include a data acquisition unit 111, a preprocessing unit 112, a model learning unit 113, and a model evaluation unit 114.

The data acquisition unit 111 can acquire data necessary for machine learning. Since a large amount of data is required for machine learning, the data acquisition unit 111 may receive image data of a plurality of existing bedsores and information about bedsores corresponding to the image data of a plurality of existing bedsores.

According to an embodiment of the present disclosure, information about bedsores may include “pressure ulcer stage information” in which a medical professional quantifies the progress of the bedsore or the severity of the bedsore in stages based on a predetermined standard. Additionally, information about bedsores may further include “beds ulcer diagnosis information.” For example, pressure ulcer diagnosis information includes information about whether the wound is severe, whether edema is severe, the amount of necrotic tissue, the type of necrotic tissue, how much epithelialization has progressed, the amount of exudate, and information about changes in the skin around the wound. It may include at least one of the above, and may include score information that scores the evaluation thereof. According to an embodiment of the present disclosure, the pressure ulcer diagnosis information may include information used in the dressing recommendation algorithm of FIGS. 8 to 10, which will be described later, for example, score information for the step-by-step questions of FIGS. 8 to 10.

The preprocessor 112 may preprocess the acquired image data of a plurality of existing bedsores so that the received data can be used for machine learning. According to an embodiment of the present disclosure, the preprocessor 112 can annotate information about pressure ulcers in pressure ulcer image data so that it can be used by the model learning unit 113, which will be described later, and the acquired image data. It can be processed into a preset format. For example, the preprocessor 112 may apply a filter to pressure ulcer image data to amplify and vectorize the data. In addition, the preprocessor 112 can adjust at least one of the size, white balance, color, contrast, or saturation of the image data so that the characteristics of the bedsore are clearly revealed, and a filter can be applied to clearly reveal the outline of the bedsore. .

Image data that has undergone preprocessing in the preprocessing unit 112 may be provided to the model learning unit 113. According to an embodiment of the present disclosure, the model learning unit 113 may learn the relationship between image data of existing bedsores and information about the corresponding bedsores. For example, the model learning unit 113 may provide pressure ulcer location information extracted from pressure ulcer image data, such as information on which part of the patient's body the pressure ulcer is located, or coordinate information indicating the area of the pressure ulcer, and pressure ulcer stage information. You can learn the relationships between Additionally, the model learning unit 113 may obtain a machine learning model that outputs information about pressure ulcers and/or pressure ulcer treatment information according to image data of existing pressure ulcers. The model learning unit 113 can learn a machine learning model using image data of existing bedsores, information about bedsores, and bedsore treatment information as learning data. Additionally, the machine learning model may be a pre-built model. For example, the machine learning model may be a model built in advance by receiving basic learning data (for example, pressure ulcer image data, pressure ulcer location information, pressure ulcer stage information, pressure ulcer diagnosis information, or pressure ulcer treatment information, etc.).

A machine learning model can be built considering the application field of the learning model, the purpose of learning, or the computer performance of the device. The machine learning model may be, for example, a model based on a neural network. For example, models such as DNN (Deep Neural Network), RNN (Recurrent Neural Network), LSTM (Long Short-Term Memory models), BRDNN (Bidirectional Recurrent Deep Neural Network), and CNN (Convolutional Neural Networks) can be used as machine learning models. However, it is not limited to this.

According to various embodiments, when there are a plurality of pre-built machine learning models, the model learning unit 113 may determine a machine learning model with a high correlation between the input learning data and the basic learning data as the machine learning model to be learned. there is. In this case, the basic learning data may be pre-classified by data type, and the machine learning model may be pre-built by data type. For example, basic learning data may be pre-classified based on various criteria, such as the location where the learning data was created, the time the learning data was created, the size of the learning data, the creator of the learning data, and the type of object in the learning data.

Additionally, the model learning unit 113 may train a machine learning model using, for example, a learning algorithm including error back-propagation or gradient descent.

Additionally, the model learning unit 113 may learn a machine learning model through, for example, supervised learning using training data as input. In addition, the model learning unit 113 creates a machine learning model through unsupervised learning, which discovers standards for situation judgment by, for example, learning the type of data needed for situation judgment without any special guidance. You can learn. Additionally, the model learning unit 113 may learn a machine learning model through, for example, reinforcement learning that uses feedback regarding whether the result of situational judgment based on learning is correct.

Additionally, when the machine learning model is learned, the model learning unit 113 may store the learned machine learning model. In this case, the model learning unit 113 may store the learned machine learning model in the memory of the electronic device including the data recognition unit 120. Alternatively, the model learning unit 113 may store the learned machine learning model in the memory of a server connected to the electronic device through a wired or wireless network.

For example, the memory in which the learned machine learning model is stored may also store commands or data related to at least one other component of the electronic device. Additionally, memory may store software and/or programs. A program may include, for example, a kernel, middleware, an application programming interface (API), and/or an application program (or “application”).

The model evaluation unit 114 inputs evaluation data into the machine learning model, and when the results output from the evaluation data do not meet a predetermined standard, the model learning unit 113 can learn again. In this case, the evaluation data may be preset data for evaluating a machine learning model.

For example, the model evaluation unit 114 determines that among the results of the machine learning model learned for the evaluation data, if the number or ratio of evaluation data for which the recognition result is incorrect exceeds a preset threshold, the model evaluation unit 114 does not meet the predetermined standard. It can be evaluated as For example, when the predetermined standard is defined as a ratio of 2%, and the learned machine learning model outputs incorrect recognition results for more than 20 evaluation data out of a total of 1000 evaluation data, the model evaluation unit 114 performs the learning The machine learning model may be evaluated as unsuitable.

Meanwhile, when there are a plurality of learned machine learning models, the model evaluation unit 114 evaluates whether each learned learning model satisfies a predetermined standard, and determines the model that satisfies the predetermined standard as the final machine learning model. You can. In this case, when there are multiple models that satisfy a predetermined standard, the model evaluation unit 114 may determine one or a predetermined number of models preset in order of highest evaluation score as the final machine learning model. Details about the model evaluation unit 114 will be described later with reference to FIG. 12.

Meanwhile, at least one of the data acquisition unit 111, the preprocessing unit 112, the model learning unit 113, and the model evaluation unit 114 in the data learning unit 110 is manufactured in the form of at least one hardware chip and is Can be mounted on the device. For example, at least one of the data acquisition unit 111, preprocessing unit 112, model learning unit 113, and model evaluation unit 114 is manufactured in the form of a dedicated hardware chip for artificial intelligence (AI). It may be manufactured as part of an existing general-purpose processor (eg, CPU or application processor) or graphics-specific processor (eg, GPU) and installed in the various electronic devices described above.

Additionally, the data acquisition unit 111, preprocessing unit 112, model learning unit 113, and model evaluation unit 114 may be mounted on one electronic device, or may be mounted on separate electronic devices. there is. For example, some of the data acquisition unit 111, preprocessing unit 112, model learning unit 113, and model evaluation unit 114 may be included in the electronic device, and the remaining part may be included in the server.

Additionally, at least one of the data acquisition unit 111, the pre-processing unit 112, the model learning unit 113, and the model evaluation unit 114 may be implemented as a software module. If at least one of the data acquisition unit 111, the preprocessing unit 112, the model learning unit 113, and the model evaluation unit 114 is implemented as a software module (or a program module including instructions), the software The module may be stored in non-transitory computer readable media that can be read by a computer. Additionally, in this case, at least one software module may be provided by an operating system (OS) or a predetermined application. Alternatively, part of at least one software module may be provided by an operating system (OS), and the remaining part may be provided by a predetermined application.

The data recognition unit 120 according to an embodiment of the present disclosure may include a data acquisition unit 121, a pre-processing unit 122, a recognition result providing unit 123, and a model updating unit 124.

The data acquisition unit 121 may obtain image data of the current patient's pressure ulcers or information about the current patient's pressure ulcers. The preprocessor 122 may preprocess the acquired data so that the acquired image data of the current patient's pressure ulcers and information related to the current patient's pressure ulcers can be used in a machine learning model. For example, the preprocessor 122 may process pressure ulcer image data with a filter to amplify and vectorize it. Additionally, the preprocessor 122 can make the outline of the bedsore clearly visible and adjust at least one of white balance, color, contrast, or saturation of the bedsore image data.

The recognition result providing unit 123 may obtain result data by applying the data preprocessed in the preprocessing unit 122 to the machine learning model as described above. The recognition result providing unit 123 may output result data in voice or text.

The model updating unit 124 may update the machine learning model based on the evaluation of the recognition result provided by the recognition result providing unit 123. For example, the model updating unit 124 may provide the recognition result provided by the recognition result providing unit 123 to the model learning unit 113, thereby allowing the model learning unit 113 to update the machine learning model. there is.

Meanwhile, at least one of the data acquisition unit 121, the pre-processing unit 122, the recognition result providing unit 123, and the model updating unit 124 in the data recognition unit 120 is manufactured in the form of at least one hardware chip. It can be mounted on an electronic device. For example, at least one of the data acquisition unit 121, the preprocessing unit 122, the recognition result providing unit 123, and the model updating unit 124 is in the form of a dedicated hardware chip for artificial intelligence (AI). It may be manufactured, or it may be manufactured as part of an existing general-purpose processor (eg, CPU or application processor) or graphics-specific processor (eg, GPU) and installed in the various electronic devices described above.

In addition, the data acquisition unit 121, the pre-processing unit 122, the recognition result providing unit 123, and the model updating unit 124 may be mounted on one electronic device, or may be mounted on separate electronic devices. It may be possible. For example, some of the data acquisition unit 121, the pre-processing unit 122, the recognition result providing unit 123, and the model updating unit 124 may be included in the electronic device, and the remaining parts may be included in the server.

Additionally, at least one of the data acquisition unit 121, the pre-processing unit 122, the recognition result providing unit 123, and the model updating unit 124 may be implemented as a software module. When at least one of the data acquisition unit 121, the preprocessing unit 122, the recognition result providing unit 123, and the model updating unit 124 is implemented as a software module (or a program module including instructions), The software module may be stored on non-transitory computer readable media. Additionally, in this case, at least one software module may be provided by an operating system (OS) or a predetermined application. Alternatively, part of at least one software module may be provided by an operating system (OS), and the remaining part may be provided by a predetermined application.

Figure 2 is a diagram showing a bedsore management system according to an embodiment of the present disclosure.

According to one embodiment of the present disclosure, the bedsore management system 200 may include an input unit 210, a bedsore management device 220, and an output unit 230. Additionally, the pressure ulcer management system 200 may further include a care integration device 240.

The input unit 210 may be included in the user terminal. The input unit 210 may acquire image data of bedsores. Additionally, the input unit 210 can obtain information about bedsores. The image data of the bedsore may be an image of the bedsore taken directly by the user using a camera. In addition, the image data of a bedsore may be a photograph of the bedsore itself, but the bedsore may also be included in the image data of a special sheet on which the patient was lying, a dressing attached to the patient's bedsore, or a patient's clothes. Pressure ulcers are different from other wounds, and the amount or color of exudate from the pressure ulcer may have a strong correlation with information about the pressure ulcer. Therefore, the image of the exudate on the patient's clothes worn by the patient with bedsores, the image of the exudate on the dressing attached to the bedsore, or the image of the exudate on the sheet on which the patient was lying can be used as information about the bedsore or pressure ulcer treatment information using a machine learning model. can be used to predict. In addition, since the image of the exudate on the patient's clothes and the image of the exudate on the sheet may differ depending on the time the patient lies down, the input unit 210 may further receive information related to the time the sheet, dressing, or patient's clothes were with the patient. . The input unit 210 can merge time-related information into the image. For example, the input unit 210 may include information related to time in the metadata of the image. Time-related information can be used in machine learning models along with video data.

Additionally, the image data of bedsores may be image data received by a user wired or wirelessly from another user terminal. Additionally, the image data of a bedsore may be image data taken together with a cotton swab to indicate the depth of the bedsore in order to more precisely measure the depth of the bedsore. Additionally, the image data of the bedsore may be image data already stored in the user terminal. Information about bedsores input to the input unit 210 may include pressure ulcer stage information and/or bedsore diagnosis information, as described above.

The input unit 210 may transmit image data of bedsores or information related to bedsores to the bedsore management device 220 via wired or wireless means. The input unit 210 may compress image data of bedsores or information related to bedsores based on a predetermined algorithm in order to reduce data capacity.

The bedsore management device 220 may be included in a user terminal or server. The bedsore management device 220 can be implemented by installing an app on a user terminal. User terminals may include tablets, laptops, smartphones, smartwatches, etc. The pressure ulcer management device 220 may perform machine learning based on data received from the input unit 210 or output result data based on an already learned machine learning model. The pressure ulcer management device 220 may output a machine learning model based on the received image data of the pressure ulcer and information related to the pressure ulcer. Additionally, the pressure ulcer management device 220 may apply the received image data of pressure ulcers to a machine learning model to output information about pressure ulcers and/or pressure ulcer treatment information.

The pressure ulcer management device 220 may transmit the generated machine learning model, information about pressure ulcers, or pressure ulcer treatment information to the output unit 230 via wired or wireless communication. The pressure ulcer management device 220 compresses the generated machine learning model, information about pressure ulcers, or pressure ulcer treatment information using a predetermined algorithm, thereby reducing the data volume required for data transmission and increasing transmission speed.

The output unit 230 may correspond to a user terminal. The output unit 230 may be included in the same user terminal as the input unit 210. The output unit 230 may apply pressure ulcer image data stored in the memory to a machine learning model to output information about pressure ulcers and/or pressure ulcer treatment information. Additionally, the output unit 230 may output information about bedsores and/or bedsore treatment information received from the bedsore management device 220. The output unit 230 can output the results to a display unit or output the results as sound through a speaker. The operation of the bedsore management device 220 will be described later with reference to FIG. 3.

The care integration device 240 may receive information about pressure ulcers from the pressure ulcer management device 220 . For example, the medical treatment integration device 240 may receive at least one of image data of a pressure ulcer, pressure ulcer location information, and information about the pressure ulcer from the pressure ulcer management device 220. The medical treatment integration device 240 may display at least one of pressure ulcer image data, pressure ulcer location information, and pressure ulcer information so that the medical staff can check at least one of pressure ulcer image data, pressure ulcer location information, and pressure ulcer information. Additionally, in order to receive responses to questions related to bedsores from medical staff, the medical treatment integration device 240 may display preset questions related to bedsores. The medical treatment integration device 240 may transmit the input response information to the pressure ulcer management device 220. In addition to the response information, the medical treatment integration device 240 may transmit information about the bedsores entered by the medical staff to the bedsore management device 220. The care integration device 240 may be located in a medical institution, such as a hospital. At this time, questions related to bedsores may be set by receiving input from an administrator or medical staff who manages the bedsore management system 200, or may be set using a preset algorithm (eg, program, etc.). In addition, questions related to pressure ulcers for which medical staff will enter responses include questions about pressure ulcer diagnostic information, such as whether the edema is severe, whether the wound is deep, what the amount of necrotic tissue is, what the amount of exudate is, and whether the fever is severe, or This may include whether there is a diagnosis of infection other than fatigue and body aches.

Figure 3 is a flowchart showing the operation of a pressure ulcer management device according to an embodiment of the present disclosure.

First, the pressure ulcer management device 220 may acquire image data of a plurality of existing pressure ulcers (S310). According to an embodiment of the present disclosure, the image data of a plurality of existing bedsores is image data obtained from bedsores of existing patients, and is obtained through the data acquisition unit 111 shown in FIG. 1 and/or the input unit 210 shown in FIG. 2. ) can be entered. Additionally, the image data of multiple existing bedsores may be images of bedsores taken by medical personnel while treating a patient. In addition, image data of multiple existing bedsores may be image data that clearly shows the area of the bedsore so that it can be used for machine learning. In addition, the image data of multiple existing bedsores may be images of the bedsores themselves, but may also be images of exudate on the patient's clothes, images of exudate on the dressing attached to the bedsore, or images of exudate on the sheet on which the patient was lying. . In addition, when receiving an image of exudate on a patient's clothes, an image of an exudate on a dressing attached to a bedsore, or an image of an exudate on a sheet on which the patient was lying, the bedsore management device 220 determines whether the sheet, dressing, or patient clothes are with the patient. Additional time-related information can be received. The bedsore management device 220 may merge time-related information with image data. For example, the bedsore management device 220 may record time-related information in metadata of image data, or transform the bit value of the time-related information into the bit value of at least one pixel and merge it into the image. If the pixel included in the image has 16 bits, the pressure ulcer management device 220 can express the time-related information in 16*n bits and convert it into a pixel value. Here, n may be a natural number. Information related to time and values converted to pixels may be proportional. Time-related information and image data can be used to create machine learning models. Users can also select image data that can be used for machine learning among a plurality of existing bedsore image data.

Additionally, according to an embodiment of the present disclosure, annotation processing may be performed on image data of a plurality of existing bedsores. For example, the user displays the area where the pressure ulcer occurs in the image data of a plurality of existing pressure ulcers using a graphical user interface (GUI) on the input unit 210 or the pressure ulcer management device 220, and displays the area where the pressure ulcer occurs as a square box. You can input pressure ulcer stage information corresponding to the image data. When a square box is designated by a user's input, the coordinates of the square box may be automatically designated by the bedsore management device 220. For example, by displaying the positions of the four vertices of a square box as x and y coordinates, the area where multiple existing pressure ulcers occur can be expressed as coordinate information, and the pressure ulcer stage information entered together when specifying the square box corresponds to this coordinate information. It can be. More detailed information regarding this will be described later with reference to FIG. 5.

Next, the pressure ulcer management device 220 may obtain information about existing pressure ulcers corresponding to image data of a plurality of existing pressure ulcers (S320).

According to an embodiment of the present disclosure, a user may obtain information about pressure ulcers corresponding to the image data of a plurality of existing pressure ulcers, along with image data of a plurality of existing pressure ulcers, through the data acquisition unit 111 shown in FIG. 1 or the data acquisition unit 111 shown in FIG. 2. It can be input into the input unit 210 shown in . Alternatively, information about pressure ulcers input through the medical treatment integration device 240 shown in FIG. 2 may be transmitted to the pressure ulcer management device 220. According to another embodiment of the present disclosure, the pressure ulcer management device 220 obtains information about pressure ulcers based on the patient's medical records corresponding to image data of existing pressure ulcers, or based on the results of existing diagnostic equipment such as laser Doppler. Thus, it is possible to obtain information about bedsores corresponding to image data of existing bedsores. The pressure ulcer management device 220 may obtain information in this way from the care integration device 240. At this time, the information about the pressure ulcers corresponding to the image data of the plurality of existing pressure ulcers may include at least one of pressure ulcer stage information, pressure ulcer diagnosis information, pressure ulcer location information in the image, and pressure ulcer treatment information. Information about bedsores may include at least one of text or numbers as an index. Information about bedsores may be a ground truth value or may be label information. Information about bedsores can be used to create machine learning models.

Next, the pressure ulcer management device 220 may acquire a machine learning model by learning the relationship between image data of a plurality of existing pressure ulcers and information about the existing pressure ulcers (S330). More details regarding this will be described later with reference to Figure 4.

Afterwards, the bedsore management device 220 may acquire bedsore image data of the current patient (S340).

According to an embodiment of the present disclosure, a user can capture a patient's bedsores using a camera included in the user terminal and transmit the pictures to the bedsore management device 220. Alternatively, the user may transmit image data already stored in the memory to the bedsore management device 220. Alternatively, the user may transmit image data received from another terminal to the bedsore management device 220.

Next, the pressure ulcer management device 220 applies a machine learning model to the current patient's pressure ulcer image data to determine information about the current patient's pressure ulcer and/or pressure ulcer treatment information (S350), and determines the pressure ulcer treatment information about the current patient's pressure ulcer. Information and/or bedsore treatment information may be output to the output unit 230 (S360).

According to an embodiment of the present disclosure, the pressure ulcer management device 220 applies the machine learning model obtained in step S330 described above to the pressure ulcer image data of the current patient to provide information about the current patient's pressure ulcer and/or pressure ulcer treatment information. You can decide. Detailed information regarding this will be described later with reference to FIG. 7.

Additionally, according to an embodiment of the present disclosure, the bedsore management device 220 may use a dressing recommendation algorithm in determining bedsore treatment information by applying a machine learning model to the current patient's bedsore image data. Detailed information regarding this will be described later along with FIGS. 8 to 11.

Figure 4 is a block diagram showing a pressure ulcer management device according to an embodiment of the present disclosure.

According to one embodiment of the present disclosure, the bedsore management device 220 may include a data learning unit 410. The data learning unit 410 of FIG. 4 may correspond to the data learning unit 110 of FIG. 1 .

According to an embodiment of the present disclosure, the data learning unit 410 may acquire image data 421 of a plurality of existing bedsores. Additionally, the data learning unit 410 may acquire information 422 about existing pressure ulcers corresponding to each of the image data of a plurality of existing pressure ulcers. Information 422 about existing bedsores may be label data. Information 422 about existing bedsores may be determined by a medical professional and input into the data learning unit 410. Alternatively, information 422 about existing bedsores may be previously stored in memory. The information 422 about existing bedsores may be an index and may include at least one of numbers or text.

The data learning unit 410 may learn the relationship between image data 421 of a plurality of existing pressure ulcers and information 422 about a plurality of existing pressure ulcers and output the information as a machine learning model 431. As described above, the information 422 regarding a plurality of existing pressure ulcers includes at least one of pressure ulcer stage information, pressure ulcer diagnosis information, location information of pressure ulcers in the image data 421, and existing pressure ulcer treatment information. can do. Here, the bedsore diagnosis information may include information used in the dressing recommendation algorithm of FIGS. 8 to 10, which will be described later. For example, pressure ulcer diagnosis information may include score information for questions at each stage of the dressing recommendation algorithm of FIGS. 8 to 10, which will be described later. The existing bedsore treatment information may be information about a treatment method corresponding to the image data 421 of the existing bedsores. The treatment method may include information about dressings or information about treatment drugs. According to an embodiment of the present disclosure, the pressure ulcer management device 220 shown in FIG. 2 may store a machine learning model 431. Additionally, the pressure ulcer management device 220 may transmit the machine learning model 431 to another pressure ulcer management device. The pressure ulcer management device 220 may apply new pressure ulcer image data to the machine learning model 431 to obtain information about pressure ulcers and/or pressure ulcer treatment information.

Additionally, according to an embodiment of the present disclosure, the machine learning model 431 shown in FIG. 4 can perform both machine learning to obtain information about bedsores and pressure ulcer treatment information. According to another embodiment of the present disclosure, the machine learning model 431 shown in FIG. 4 may include at least one of a machine learning model for acquiring information about pressure ulcers and a machine learning model for acquiring pressure ulcer treatment information. You can.

According to an embodiment of the present disclosure, the bedsore management device 220 may apply the machine learning model 431 to the bedsore image data of the current patient to determine information about the current patient's bedsores or pressure ulcer treatment information. The process by which the pressure ulcer management device 220 acquires information about pressure ulcers and/or pressure ulcer treatment information using the machine learning model 431 will be described later with reference to FIG. 7 .

According to another embodiment of the present disclosure, the bedsore management device 220 may further use sensor information as well as image data. For example, an olfactory sensor or dust sensor may be attached to the patient's side. The bedsore management device 220 may periodically obtain sensor information from an olfactory sensor or a dust sensor. Sensor information can be displayed as bit values. The bedsore management device 220 may merge bit values of sensor information into image data. For example, the pressure ulcer management device 220 may transform the bit value of the sensor information into the bit value of at least one pixel and merge it into the image. If the pixels included in the image data are 16 bits, sensor information can be expressed in 16*n bits. Here, n may be a natural number. The bedsore management device 220 can perform machine learning on image data of existing bedsores including sensor information, information on existing bedsores, or pressure ulcer treatment information. Through this, the bedsore management device 220 can acquire a machine learning model. Additionally, the bedsore management device 220 may determine information about the current patient's bedsores or pressure ulcer treatment information using the acquired machine learning model. That is, the bedsore management device 220 can determine information about the current patient's bedsores or pressure ulcer treatment information by applying the machine learning model 431 to the bedsore image data of the current patient. Here, the current patient's bedsore image data may be merged with sensor information using the method already described above.

Hereinafter, a machine learning network for the data learning unit 410 shown in FIG. 4 to acquire the machine learning model 431 will be described.

The data learning unit 410 shown in FIG. 4 can acquire a machine learning model 431 by learning the relationship between image data of a plurality of existing pressure ulcers and information about existing pressure ulcers using a learning network using a neural network.

According to an embodiment of the present disclosure, the network for the data learning unit 410 to learn the relationship between image data of a plurality of existing pressure ulcers and information about existing pressure ulcers is a CNN (Convolutional Neural Network). It can be done, but is not limited to this.

According to another embodiment of the present disclosure, the data learning unit 410 determines the relationship between image data of a plurality of existing pressure ulcers and information about existing pressure ulcers using R-CNN (Region-based Convolutional Neural Network). You can obtain a machine learning model (431) by learning.

According to another embodiment of the present disclosure, the data learning unit 410 learns the relationship between image data of a plurality of existing pressure ulcers and information about existing pressure ulcers with Faster R-CNN to obtain a machine learning model 431. You can.

As is widely known, there are various models of neural networks that are the basis of machine learning models. For example, representative neural networks include CNN and models improved from CNN. Among them, there is a technology called R-CNN that classifies objects included in image data using a deep neural network.

Basically, CNN is a network consisting of several layers, including a convolutional layer and a pooling layer. Filtering is performed on the input image through a convolutional layer, and a feature map is extracted as a result. The feature map is used as input to the next layer and processed continuously for each layer. CNN outputs classification or analysis results for the input image based on the feature map for the input image.

R-CNN is a model that connects a CNN that performs object analysis and an algorithm (region proposal algorithm) that suggests a region where an object will exist in the image. R-CNN determines a rectangular candidate area using an algorithm such as selective search in the input image. R-CNN cuts candidate regions to the same size, extracts features for each region, and classifies objects.

On the other hand, Faster R-CNN is fast because it uses a method of cropping the feature map rather than cropping the image itself. Faster R-CNN uses RPN (region proposal network) to determine candidate regions. Specifically, in order to recognize an object included in the image data, the object area in the image data is designated using RPN so that areas other than the object required for recognition can be removed.

According to an embodiment of the present disclosure, the data learning unit 410 performs pre-processing including annotations on the acquired image data 421 of a plurality of existing pressure ulcers, and extracts image features for the vectorized image data through the pre-processing. , the location information of the pressure ulcer occurrence area can be extracted through the RPN stage and the detector stage. Through this process, the data learning unit 410 can learn the relationship between pressure ulcer location information and pressure ulcer stage information. Detailed information regarding this will be described later along with FIGS. 5 and 6.

Figure 5 is a diagram illustrating an example of an interface and process for annotating image data of existing bedsores according to an embodiment of the present disclosure.

According to an embodiment of the present disclosure, the pressure ulcer management device 220 may perform annotation processing to display pressure ulcer location information and pressure ulcer stage information on image data of a plurality of existing pressure ulcers. Afterwards, the pressure ulcer management device 220 can infer pressure ulcer location information and pressure ulcer stage information from the pressure ulcer image data by learning the annotated data. Below, an embodiment of a process for the pressure ulcer management device 220 to perform annotation processing on image data of a plurality of existing pressure ulcers will be described.

As shown in FIG. 5, original image data (image) and a graphical user interface (GUI) of a plurality of existing bedsores may be displayed on the user terminal. Coordinate information of the pressure ulcer occurrence location may be input to the input unit 210 by the user designating the pressure ulcer occurrence area on the image data of the existing pressure ulcer through the GUI. Additionally, users can input information about existing bedsores through the GUI. For example, as shown in FIG. 5, the user can input pressure ulcer stage information of existing pressure ulcers as information about existing pressure ulcers through the GUI. The input information can be obtained from the data acquisition unit 111 shown in FIG. 1. Additionally, although not shown in FIG. 5, the user may input additional diagnostic information regarding existing bedsores.

As shown in FIG. 5, the user can display the area where the pressure ulcer occurs in the image data of a plurality of existing pressure ulcers using a square box through the GUI. When a square box is input through the GUI, the coordinates of X1, Y1, X2, and Y2 corresponding to the four vertices of the square box can be automatically specified, as shown in FIG. 5. This can be done by the data acquisition unit 111 or the preprocessing unit 112 shown in FIG. 1.

Additionally, according to an embodiment of the present disclosure, the preprocessor 112 shown in FIG. 1 may convert the file format of image data of existing bedsores acquired by the data acquisition unit 111. For example, as shown in FIG. 5, when the format of the image data acquired by the data acquisition unit 111 is JPG or PNG, the preprocessor 112 may convert the data into XLSX format or JSON format. Data converted to XLSX format can be used as inspection data, and data converted to JSON format can be used as learning data for machine learning.

Additionally, according to an embodiment of the present disclosure, the accuracy of annotation processing can be improved by having at least one professional medical professional provide feedback on data for inspection. Specifically, at least one medical professional determines whether the image data of the existing pressure ulcer and the information about the corresponding pressure ulcer are appropriately matched and provides feedback, and the user re-enters the updated information applying the feedback through the GUI. You can.

Referring again to the above-mentioned FIG. 4, the image data 421 of a plurality of existing pressure ulcers and the information 422 about the existing pressure ulcers are annotated as above, constructed as learning data, and input into the data learning unit 410. It can be used for machine learning after certain data preprocessing. Detailed information regarding this will be described later with reference to FIG. 6.

Furthermore, although not shown in FIG. 4, the preprocessing unit 112 provides pressure ulcer diagnosis information, pressure ulcer assessment score (Braden Scale) and albumin level, wound bed condition and amount of exudate, characteristics of each dressing material, and wound healing. By annotating at least one of the features according to the presence and type of healing, various features related to pressure ulcers can be machine learned.

Figure 6 is a diagram illustrating an example of an interface and process for machine learning of image data of a plurality of existing pressure ulcers according to an embodiment of the present disclosure.

According to an embodiment of the present disclosure, image data of a plurality of existing pressure ulcers converted into learning data through annotation processing, as described above in FIG. 5, may be subjected to a series of processes for machine learning in the data learning unit 410. there is. For example, as shown in FIG. 6, first, the pre-processing unit 112 of the data learning unit 410 performs a pre-processing step, an image feature extraction step, an RPN step, and a detector step on the received image data of a plurality of existing pressure ulcers. Through this, the location information and stage information of image data of a plurality of existing bedsores can be output.

According to an embodiment of the present disclosure, the preprocessor 112 may apply an image filter to image data of a plurality of existing bedsores to amplify the image data and process it into a vector form so that it can be used for machine learning. For example, the preprocessor 112 may amplify the image data of a plurality of existing pressure ulcers by applying at least one of the bilateral, Gaussian, and histogram equalization filters to the image data of the existing pressure ulcers. there is.

Next, as shown in FIG. 6, the preprocessor 112 performs normalization, resizing, flipping, padding, scaling, and positioning on the amplified image data. Pressure ulcer image data can be vectorized by performing one or more of the following processing: translation, rotation, perspective transformation, noise adjustment, lighting condition change, and background and style change.

Here, normalizing image data means limiting the range of data to a user-desired range in order to reduce scale deviation of data characteristics. Resizing is used to create an image of an object that has a similar appearance but looks different depending on the viewing angle by creating an image with a changed appearance by adjusting the proportion of the object within the object image. Flipping is used to create an image of an object by changing the top and bottom or left and right sides of the object to create an image of an object that has the same appearance but looks different depending on the viewing angle and direction. Scaling is used to create object images of various sizes by modifying the size of the object image, and to create object images that have the same appearance but look different depending on the viewing angle and distance. Position change is used to change the background and position of the object image to create an object image that has a similar appearance but appears different depending on the viewing position. Rotation is used to create an image of an object that has a similar appearance but appears different depending on the viewing angle by rotating the image to create images from various angles. Perspective conversion is used to create an image of an object that has a similar appearance but looks different depending on the viewing angle by creating an image by adjusting the angle of the object's visible three-dimensional (X-, Y-, and Z-axis) directions. Noise adjustment is used to create image data that has the same appearance but divides the degree of external texture of the object into several levels. Illuminance change is used to create image data by changing the brightness of the object image, including various colors and shadows depending on the brightness of the light.

Next, as shown in FIG. 6, the model learning unit 113 uses a neural network for the image data vectorized by the preprocessor 112 to create a feature map of the pressure ulcer image data through several scales. It can be obtained. Next, the model learning unit 113 can apply RPN (region proposal network) to the obtained feature map to extract information about the area to be recognized, for example, the area where the pressure ulcer occurs, from the image data of the existing pressure ulcer. In the step of applying RPN, binary classification can be performed on an anchor box consisting of a predetermined horizontal and vertical ratio based on a specific position in the image using the feature map of the acquired image data.

Next, as shown in FIG. 6, the model learning unit 113 can apply the detector after applying the RPN. In the step of applying the detector, the RPN results and the feature map of the image data are used to obtain information about the location of a plurality of existing pressure ulcers, for example, location coordinate information of the pressure ulcer occurrence area or depth information of the pressure ulcer, and the corresponding existing pressure ulcer. Information about bedsores, for example, pressure ulcer stage information, can be learned.

In addition, according to an embodiment of the present disclosure, the model learning unit 113 calculates the final loss value by adding the RPN loss and the detector loss, and the parameters of the machine learning model are the final loss value. can be updated in the direction of reducing.

In addition, according to an embodiment of the present disclosure, the model learning unit 113 may use a technology that combines SOTA (State of the art) object recognition technology in extracting image features from image data, and COCO (Common Objects in Context) dataset can also be used.

Figure 7 is a block diagram showing a pressure ulcer management device according to an embodiment of the present disclosure.

According to one embodiment of the present disclosure, the bedsore management device 220 shown in FIG. 2 may include the data recognition unit 120 of FIG. 1 . The data recognition unit 120 shown in FIG. 1 may obtain result data by applying input data to a previously obtained machine learning model 710. Specifically, the bedsore management device 220 may obtain information 730 about the bedsores of the current patient by applying the bedsore image data 720 of the current patient to the machine learning model 710.

The current patient's bedsore image data acquired by the bedsore management device 220 may be a photograph of the bedsore itself, but may also be an image of the exudate on the patient's clothes, an image of the exudate on the dressing attached to the bedsore, or on the sheet on which the patient was lying. It may be an image of exudate. When receiving an image of exudate on a patient's clothing, an image of exudate on a dressing attached to a bedsore, or an image of exudate on a sheet on which the patient was lying, the bedsore management device 220 merges time-related information with the image data. You can. For example, the bedsore management device 220 may record time-related information in metadata of image data, or transform the bit value of the time-related information into the bit value of at least one pixel and merge it into the image. Since the process of merging image data and time-related information has already been described, redundant explanation will be omitted.

Additionally, the bedsore management device 220 may obtain additional bedsore treatment information 740 by applying the current patient's bedsore image data 720 to the machine learning model 710. The machine learning model 710 of FIG. 7 may correspond to the machine learning model 431 of FIG. 4.

If the machine learning model 431 in FIG. 4 machine-learned the relationship between image data of existing pressure ulcers and information about existing pressure ulcers, the machine learning model 431 in FIG. 7 is based on the current patient's pressure ulcer image data. You can print out information about bedsores or bedsore treatment information. This is because the information about existing bedsores includes at least one of pressure ulcer stage information of existing bedsores, bedsore diagnosis information, location information of bedsores in the image data 421, and existing bedsore treatment information.

Here, the information 730 about the current patient's pressure ulcers, which the machine learning model 710 obtains by learning the current patient's pressure ulcer image data 720, may include pressure ulcer stage information and/or diagnosis information. The current patient's pressure ulcer diagnosis information may include information used in the dressing recommendation algorithm of FIGS. 8 to 10, which will be described later. For example, the current patient's pressure ulcer diagnosis information may include score information for questions at each stage of the dressing recommendation algorithm of FIGS. 8 to 10, which will be described later.

According to an embodiment of the present disclosure, the pressure ulcer management device 220 may obtain various information about pressure ulcers by applying the current patient's pressure ulcer image data 720 to the machine learning model 710. For example, the pressure ulcer management device 220 may apply the current patient's pressure ulcer image data 720 to the machine learning model 710 to obtain pressure ulcer location information and pressure ulcer stage information. In addition, the pressure ulcer management device 220 applies the current patient's pressure ulcer image data 720 to the machine learning model 710 to obtain diagnostic information about the current pressure ulcer used in the dressing recommendation algorithm described later in FIGS. 8 to 10. You can.

Furthermore, the bedsore management device 220 may obtain pressure ulcer treatment information 740 by applying the acquired pressure ulcer information 730 to the algorithm mentioned in FIGS. 8 to 10 .

However, it is not limited to this, and the pressure ulcer management device 220 can directly generate pressure ulcer treatment information by applying the current patient's pressure ulcer image data to the machine learning model 710.

Additionally, diagnostic information that must be entered directly by the user or medical professional, such as whether the patient currently has a high fever and whether there is a diagnosis of infection other than fatigue and body aches, can be input into the pressure ulcer management device 220. , Additional information input in this way can be used to determine bedsore treatment information 740.

Additionally, the bedsore management device 220 can obtain information about the current patient's bedsores or pressure ulcer treatment information by applying the current patient's bedsore image data, which includes the current patient's sensor information, to a machine learning model. As already described, the sensor information may be information obtained from an olfactory sensor or a dust sensor. Sensor information can be merged with the patient's current pressure ulcer image data. Since the merging process has already been described, redundant explanation will be omitted.

According to an embodiment of the present disclosure, the bedsore treatment information 740 may include dressing information necessary for bedsore treatment. A dressing for treating bedsores may be determined by a dressing recommendation algorithm. Specific details regarding the dressing recommendation algorithm will be described later with reference to FIGS. 8 to 10.

In addition, the dressing recommendation algorithm in FIGS. 8 to 10, which will be described later, is an embodiment according to the present disclosure, and such a dressing recommendation algorithm may be determined through machine learning. This will be described later with reference to FIG. 11.

8 to 10 are flowcharts showing a dressing recommendation algorithm according to an embodiment of the present disclosure.

Referring to FIGS. 8 to 10 , the bedsore management device 220 uses pressure ulcer diagnosis information to determine whether the patient has a severe high fever (811).

If the patient's high fever is 37.8°C or higher, the bedsore management device 220 requests a hospital visit through the user terminal or requests a home visit professional nurse through the care integration device 240 (813).

If the patient's high fever is less than 37.8°C, the pressure ulcer management device 220 checks the response as to whether the wound is severe obtained from the pressure ulcer management device 220 and/or the care integration device 240 (817). Currently, diagnostic information regarding bedsores may include response information regarding whether the wound is severe. The bedsore management device 220 can obtain response information about whether the wound is severe based on the machine learning model 710. That is, the bedsore management device 220 can apply the current patient's bedsore image data to the machine learning model 710 to obtain response information about whether the wound is severe.

If the response to whether the wound is severe is 5 points, the pressure ulcer management device 220 sets the dressing treatment to a hydrofiber dressing (819). The subsequent process will be described in step 835.

If the response to whether the wound is severe is less than 5 points, the pressure ulcer management device 220 checks the response to whether the edema is severe obtained from the pressure ulcer management device 220 and/or the care integration device 240 (821). Currently, diagnostic information regarding pressure ulcers may include response information regarding whether the edema is severe. The pressure ulcer management device 220 can obtain response information about whether the edema is severe based on the machine learning model 710. That is, the bedsore management device 220 can apply the current patient's bedsore image data to the machine learning model 710 to obtain response information about whether the edema is severe.

If the response to whether the edema is severe is 5 or more, the pressure ulcer management device 220 requests a hospital visit through the user terminal or requests a home visit professional nurse through the care integration device 240 (823).

If the response to whether the edema is severe is 3 or more points but less than 5 points, the bedsore management device 220 sets the dressing treatment to a hydrofiber dressing (825). The subsequent process will be described in step 835.

If the response to whether the edema is severe is less than 3 points, the pressure ulcer management device 220 checks the response to the amount of necrotic tissue obtained from the pressure ulcer management device 220 and/or the care integration device 240 ( 827). Current diagnostic information regarding pressure ulcers may include response information regarding the amount of necrotic tissue. The bedsore management device 220 can obtain response information about the amount of necrotic tissue based on the machine learning model 710. That is, the bedsore management device 220 can obtain response information about the amount of necrotic tissue by applying the current patient's bedsore image data to the machine learning model 710.

If the response to the question about the amount of necrotic tissue is 4 or more, the pressure ulcer management device 220 requests a hospital visit through the user terminal or requests a home visit professional nurse through the medical treatment integration device 240 (829).

If the response to the question about the amount of necrotic tissue is less than 4 points, the bedsore management device 220 uses the bedsore diagnosis information to check whether there is an infection diagnosis excluding high fever, fatigue, and body aches (831).

If there is an infection diagnosis other than high fever, fatigue, and body aches, the bedsore management device 220 sets the dressing treatment to a hydrofiber dressing (833).

The pressure ulcer management device 220 confirms a response about the amount of exudate obtained from the pressure ulcer management device 220 and/or the care integration device 240 (835). Currently, diagnostic information regarding pressure ulcers may include response information regarding the amount of exudate. The bedsore management device 220 can obtain response information about the amount of exudate based on the machine learning model 710. That is, the pressure ulcer management device 220 can obtain response information about the amount of exudate by applying the current patient's pressure ulcer image data to the machine learning model 710.

If the response to the amount of exudate is 4 or more, the pressure ulcer management device 220 checks the response to the degree to which epithelialization has progressed obtained from the pressure ulcer management device 220 and/or the care integration device 240. Do it (837). Current diagnostic information regarding pressure ulcers may include response information regarding the extent to which epithelialization has progressed. The pressure ulcer management device 220 can obtain response information about the extent to which epithelialization has progressed based on the machine learning model 710. That is, the pressure ulcer management device 220 can obtain response information about the extent to which epithelialization has progressed by applying the current patient's pressure ulcer image data to the machine learning model 710.

If the response to the extent to which epithelialization has progressed is less than 3 points, the pressure ulcer management device 220 sets the dressing treatment to a sterile, adhesive dressing (839).

If the response to the extent to which epithelialization has progressed is 3 or more, the bedsore management device 220 sets the dressing treatment to a foam dressing (841).

If the response to the amount of exudate is greater than 1 point and less than or equal to 3 points, the bedsore management device 220 checks the response to the amount of necrotic tissue included in the reading result data (843).

If the response to the amount of necrotic tissue is less than 3 points, the pressure ulcer management device 220 provides a response to the degree to which epithelialization has progressed obtained from the pressure ulcer management device 220 and/or the care integration device 240. Confirm (845).

Thereafter, the proceeding process will be described in step 853.

If the response to the amount of necrotic tissue is 3 or more, the pressure ulcer management device 220 checks the response to the amount of exudate obtained from the pressure ulcer management device 220 and/or the treatment integration device 240. Do (847).

If the response to the amount of exudate is 3, the bedsore management device 220 sets the dressing treatment agent to a sterile, adhesive dressing (849).

If the response to the amount of exudate is 2, the pressure ulcer management device 220 checks the response to the degree to which epithelialization has progressed obtained from the pressure ulcer management device 220 and/or the treatment integration device 240. (851). Thereafter, the proceeding process will be described in step 853.

The pressure ulcer management device 220 confirms a response regarding the extent to which epithelialization has progressed obtained from the pressure ulcer management device 220 and/or the care integration device 240 (853).

If the response to the extent to which epithelialization has progressed is less than 3 points, the bedsore management device 220 sets the dressing treatment to a hydrocolloid dressing (855).

If the response to the extent to which epithelialization has progressed is 3 or more, the bedsore management device 220 sets the dressing treatment to a foam dressing (857).

Unless there is a diagnosis of infection other than high fever, fatigue, and body aches, the pressure ulcer management device 220 confirms a response as to the amount of exudate obtained from the pressure ulcer management device 220 and/or the care integration device 240 ( 859).

If the response to the amount of exudate is 1 point, the pressure ulcer management device 220 checks the response to whether the wound is deep obtained from the pressure ulcer management device 220 and/or the care integration device 240 (861). . Currently, diagnostic information regarding pressure ulcers may include response information regarding whether the wound is deep. The bedsore management device 220 can obtain response information about whether the wound is deep based on the machine learning model 710. That is, the bedsore management device 220 can apply the current patient's bedsore image data to the machine learning model 710 to obtain response information about whether the wound is deep.

If the response to whether the wound is deep is 3 points or less, the pressure ulcer management device 220 sets the dressing treatment to a hydrocolloid dressing (863).

If the response to whether the wound is deep is 4 points, the bedsore management device 220 sets the dressing treatment to hydrofiber dressing and hydrocolloid dressing (865).

If the response to the amount of exudate is greater than 1 point and less than or equal to 3 points, the pressure ulcer management device 220 determines the amount of necrotic tissue obtained in the pressure ulcer management device 220 and/or the care integration device 240. Check the response (867).

If the response to the amount of necrotic tissue is less than 3 points, the pressure ulcer management device 220 checks the response to the question of whether the wound is deep obtained from the pressure ulcer management device 220 and/or the care integration device 240 ( 869).

If the response to whether the wound is deep is 1 point, the pressure ulcer management device 220 sets the dressing treatment to a hydrocolloid dressing (871).

If the response to whether the wound is deep is greater than 1 point and less than or equal to 2 points, the bedsore management device 220 sets the dressing treatment agent to a hydrocolloid dressing (873).

If the response to whether the wound is deep is 3 or more, the pressure ulcer management device 220 sets the dressing treatment to a hydrofiber dressing (875).

After setting the hydrofiber dressing, the pressure ulcer management device 220 checks 877 for a response from the pressure ulcer management device 220 and/or the care integration device 240 regarding the amount of exudate obtained.

If the response to the amount of exudate is 2, the pressure ulcer management device 220 checks the response to the degree to which epithelialization has progressed obtained from the pressure ulcer management device 220 and/or the treatment integration device 240. (879).

If the response to the extent to which epithelialization has progressed is less than 3 points, the bedsore management device 220 sets the dressing treatment to a hydrocolloid dressing (881).

If the response to the extent to which epithelialization has progressed is 3 or more, the bedsore management device 220 sets the dressing treatment to a foam dressing (883).

If the response to the amount of exudate is 3, the bedsore management device 220 sets the dressing treatment agent to a sterile, adhesive dressing (885).

If the response to the amount of necrotic tissue is 3 or more, the pressure ulcer management device 220 checks the response to whether the wound is deep obtained from the pressure ulcer management device 220 and/or the care integration device 240 ( 887).

If the response to whether the wound is deep is less than 2 points, the pressure ulcer management device 220 checks the response to the amount of exudate obtained from the pressure ulcer management device 220 and/or the care integration device 240 (889). ).

If the response to the amount of exudate is 2, the pressure ulcer management device 220 checks the response to the degree to which epithelialization has progressed obtained from the pressure ulcer management device 220 and/or the treatment integration device 240. (891).

If the response to the extent to which epithelialization has progressed is less than 3 points, the bedsore management device 220 sets the dressing treatment to a hydrocolloid dressing (893).

If the response to the extent to which epithelialization has progressed is 3 or more, the pressure ulcer management device 220 sets the dressing treatment to a foam dressing (895).

If the response to the amount of exudate is 3, the bedsore management device 220 sets the dressing treatment agent to a sterile, adhesive dressing (897).

If the response to whether the wound is deep is 2 or more points, the pressure ulcer management device 220 checks the response to the amount of exudate obtained from the pressure ulcer management device 220 and/or the care integration device 240 (899) ).

If the response to the amount of exudate is 2, the bedsore management device 220 sets the dressing treatment to a hydrocolloid dressing and a sterile, adhesive dressing (901).

If the response to the amount of exudate is 3, the bedsore management device 220 sets the dressing treatment to a hydrofiber dressing and a sterile, adhesive dressing (903).

If the response to the amount of exudate is 4, the pressure ulcer management device 220 checks the response to the question of whether the wound is deep obtained from the pressure ulcer management device 220 and/or the care integration device 240 (905). .

If the response to whether the wound is deep is less than 2 points, the pressure ulcer management device 220 checks the response to the question What is the change in the skin around the wound obtained from the pressure ulcer management device 220 and/or the care integration device 240? Do it (907). Current diagnostic information regarding pressure ulcers may include response information regarding changes in the skin around the wound. The pressure ulcer management device 220 can obtain response information about changes in the skin around the wound based on the machine learning model 710. That is, the bedsore management device 220 can obtain response information about changes in the skin around the wound by applying the current patient's bedsore image data to the machine learning model 710.

If the response to the question about changes in the skin around the wound is 1, the bedsore management device 220 sets the dressing treatment to a sterile, adhesive dressing (909).

If the response to the question about changes in the skin around the wound is 2 or more, the bedsore management device 220 sets the dressing treatment to a hydrofiber dressing (911).

After setting up the hydrofiber dressing, the pressure ulcer management device 220 checks 913 for a response to the extent of epithelialization obtained from the pressure ulcer management device 220 and/or the care integration device 240.

If the response to the extent to which epithelialization has progressed is less than 3 points, the bedsore management device 220 sets the dressing treatment to a sterile, adhesive dressing (915).

If the response to the extent to which epithelialization has progressed is 3 or more, the bedsore management device 220 sets the dressing treatment to a foam dressing (917).

If the response to whether the wound is deep is 2 or more points, the pressure ulcer management device 220 checks the response to the amount of necrotic tissue obtained from the pressure ulcer management device 220 and/or the treatment integration device 240 ( 919).

If the response regarding the amount of necrotic tissue is less than 3, the pressure ulcer management device 220 returns to step 907 and determines the amount of skin around the wound obtained from the pressure ulcer management device 220 and/or care integration device 240. Check the response to what is the change (921). Thereafter, the bedsore management device 220 performs the process following step 907.

If the answer to the question about the amount of necrotic tissue is 3 or more, the pressure ulcer management device 220 determines the change in the skin around the wound obtained from the pressure ulcer management device 220 and/or the care integration device 240. Check the response (923).

If the response to the question about changes in the skin around the wound is 1, the bedsore management device 220 sets the dressing treatment to a sterile, adhesive dressing (925).

If the response to the question about changes in the skin around the wound is 2 or more, the pressure ulcer management device 220 sets the dressing treatment to a hydrofiber dressing (927).

If the response to the question about changes in the skin around the wound is 2 or more, the bedsore management device 220 additionally sets a sterile, adhesive dressing in addition to the hydrofiber dressing (929).

If the response regarding the amount of exudate is 5 points, the bedsore management device 220 sets the dressing treatment agent to a hydrofiber dressing (929).

Figure 11 shows an example of the operation flow of a bedsore management device according to an embodiment of the present disclosure.

As described above, the pressure ulcer management device according to an embodiment of the present disclosure is capable of extracting characteristic information of the image data of a plurality of existing pressure ulcers by machine learning the relationship between the image data of a plurality of existing pressure ulcers and information about the plurality of existing pressure ulcers. You can. For example, the characteristic information of the image data of a plurality of existing bedsores may include location information indicating the area where the bedsores occur and pressure ulcer stage information. Furthermore, information about bedsores may include diagnostic information corresponding to image data of bedsores. The diagnostic information may include subdivided classes of dressing methods. For example, the diagnostic information may include at least one of the degree of exudate, presence of granulation tissue, internal exposure, presence of infection, presence of dryness, and contraindications.

The pressure ulcer management device can apply a deep decision tree boosting model to the characteristic information and diagnosis information of the existing pressure ulcer image data as above in order to select the optimal dressing recommendation algorithm.

Ensemble is a treadmill technique that combines multiple decision trees to achieve better performance than a single decision tree. In the ensemble technique, the accuracy of machine learning models can be improved by combining multiple weak classifiers to create a strong classifier. Recently, ensemble techniques have shown innovative performance in machine learning fields such as computer vision, speech recognition, natural language processing, and signal processing. There are various ensemble techniques, including boosting and bagging. The deep decision tree boosting model is a model that applies an ensemble learning method called boosting to the basic decision tree. It is an algorithm that can determine the optimal deep decision tree through machine learning that can assign weights while optimizing the loss value. am.

According to an embodiment of the present disclosure, as shown in FIG. 11, diagnostic information subdivided into classes in relation to the characteristic information of the image data of pressure ulcers and the dressing method can be input as input to the deep decision tree boosting algorithm. there is. The input information goes through a mapping module and is output as a decision tree, goes through a boosting step to optimize the loss, and then goes through an ensemble learning step to perform learning using weights, resulting in an optimal decision. The entire tree can be determined.

The pressure ulcer management device according to an embodiment of the present disclosure can determine the optimal dressing method by applying the deep decision tree boosting model as described above. Furthermore, the suitability of the determined decision tree can be inspected or consulted by at least one medical expert. Afterwards, by adding information about the results of inspection or consultation, the updated information can be re-entered into the deep decision tree boosting algorithm, and in this way, the accuracy of dressing recommendations can be continuously improved by reflecting expert feedback.

Figure 12 shows an example of an evaluation process of a machine learning model according to an embodiment of the present disclosure.

The model evaluation unit 114 shown in FIG. 1 performs analysis, evaluation, and processing of the causes of errors or reduced efficiency through the process shown in FIG. 12 to improve the learning and inference speed and accuracy of the machine learning model. You can.

According to an embodiment of the present disclosure, model parameters, hyper-parameters, GPU (Graphic Processing Unit), and 16-bit (bits) operations are used to improve the speed, inference speed, and accuracy of machine learning. At least one factor can be adjusted to be optimized.

Model parameters are parameters obtained from data inside a machine learning model. They are not parameters entered by the user, but are values automatically determined by modeling. Hyperparameters are parameters that must be set in advance before performing model training, and are different from the parameters of machine learning models learned through training. The performance of a machine learning model can vary greatly depending on what value the hyperparameter is set to.

The model evaluation unit 114 may transmit a command to the model learning unit 113 to perform distributed GPU training to optimize multi-GPU learning.

Additionally, the model evaluation unit 114 may transmit a command to the model learning unit 113 to perform mixed precision training in order to reduce the proportion of memory required for machine learning and shorten the time required for machine learning. there is.

Additionally, in order to increase the efficiency of the RPN step, the model evaluation unit 114 may analyze the annotated data and transmit information on the size and ratio of the optimized RPN anchor box to the model learning unit 113.

According to an embodiment of the present disclosure, the model evaluation unit 114 may find and remove incorrect learning data that reduces the accuracy of learning in order to increase reliability and improve accuracy of machine learning data.

For example, when the pressure ulcer stage is incorrectly evaluated from the image data of the pressure ulcer, for example, when a stage 1 pressure ulcer is incorrectly classified as a stage 6, the model evaluation unit 114 generates image features for use in cause analysis from the image data. By extracting vectors and calculating the cosine similarity with the image feature vectors of the learned dataset, the top N learning data with the highest similarity within the dataset are detected, and information on the pressure ulcer stage of the corresponding image data is detected. It can be determined that the annotation processing for was incorrect. Next, the model evaluation unit 114 may perform annotation processing to correct pressure ulcer stage information or remove incorrectly annotated data from the dataset. This allows the dataset to be optimized and improves the accuracy and reliability of machine learning.

Figure 13 is a flowchart for explaining a method of operating a bedsore management device according to an embodiment of the present disclosure.

The bedsore management device 220 may further perform the following processes in addition to the steps of FIG. 3. More specifically, the pressure ulcer management device 220 may perform the process of FIG. 13 before the steps of FIG. 3 are performed.

Recently, there have been cases where home treatment has been recommended, but since it is difficult to receive help from a professional medical professional during home treatment, there may be cases where a user other than a professional medical professional takes care of the patient. The pressure ulcer management device 220 can help improve the patient's health by simply informing the patient of the patient's condition so that appropriate measures can be taken. Early bedsores are not easy for the general public to distinguish, and it is easy to take them lightly because they only appear to be slightly swollen. In this way, when the user suspects that the patient has a bedsore but is not sure about it, the user can receive help from the bedsore management device 220. If the bedsore is severe, the user can naturally tell that the patient has a bedsore. However, if the bedsore is not severe, it may not be easy for the user to judge. As with all diseases, prevention is important, so the pressure ulcer management device 220 according to the present disclosure explains the process of diagnosing pressure ulcers in the early stages. The user can improve the patient's health by checking the results of the bedsore management device 220 and responding to the initial bedsores.

The user may input a request to perform a simple pressure ulcer diagnosis into the pressure ulcer management device 220. The bedsore management device 220 may perform the operation of the bedsore management device 220 of FIG. 13 based on the user's input. Alternatively, the bedsore management device 220 may automatically perform the operation of FIG. 13 at a predetermined cycle. The bedsore management device 220 may transmit a message to the user asking whether the patient has a suspected area of bedsores. The bedsore management device 220 may output a message indicating to photograph an area suspected of having a bedsore. More specifically, the bedsore management device 220 may output a message indicating that the patient should be placed in a prone position. The user can place the patient in a prone position as shown in the bedsore management device 220. The user can check whether there are bedsores on the user's back, head, buttocks, back of the thigh, back of the calf, or heel of the foot.

The bedsore management device 220 may perform step 1310 of capturing a first image of the area where the patient's bedsore is suspected. The user may not be an expert in pressure ulcers. Therefore, you may not know whether what you are photographing is a bedsore or not. The user can take pictures of the area believed to be a bedsore using the user terminal included in the bedsore management device 220 owned by the user. The captured image may be the first image.

The first image may be captured with the area suspected to be a bedsore filled. More specifically, the first image may be captured using the selfie camera of the user terminal included in the bedsore management device 220. The bedsore management device 220 may output red light on the display unit. Users can place the selfie camera at a certain distance from the patient's suspected bedsore area and take pictures. Since red light is emitted from the display unit, the first image captured by the selfie camera may be captured so that the red color stands out. The certain distance may be 5 cm to 20 cm or less. Because the display's red light must shine on the area suspected of having a bedsore, the distance between the patient's skin and the camera lens may need to be close. However, it is not limited to this, and the bedsore management device 220 may output blue or green light on the display unit. Users can place the selfie camera at a certain distance from the patient's suspected bedsore area and take pictures. Since the display unit emits blue or green light, the red part of the skin may appear dark in the first image captured with the selfie camera. This is because there is no red color in the light emitted from the display. The certain distance may be 5 cm to 20 cm or less. Because the display's blue or green light must be illuminated on the area where pressure ulcers are suspected, the distance between the patient's skin and the camera lens may need to be close.

The pressure ulcer management device 220 may perform a step 1320 of capturing a second image of the area where a pressure ulcer of the patient is suspected after a predetermined time period after capturing the first image. After step 1310, the pressure ulcer management device 220 may output a message indicating that the patient will be maintained in the same posture as in the first image for a predetermined period of time. The predetermined time may be between 5 minutes and 15 minutes. The predetermined time is not long, so it can be a time that allows the patient to quickly check whether it is a bedsore or not without feeling uncomfortable. Even normal skin can become red when pressed. In the case of bedsores, it may take a long time for the skin to return to its normal color. Accordingly, the bedsore management device 220 can determine whether it is a bedsore by capturing the second image after a predetermined time and comparing it with the first image.

After a predetermined time has elapsed, the user can capture the same area as the area where the first image was captured using the second image. The bedsore management device 220 may acquire a second image. The pressure ulcer management device 220 outputs a message indicating to capture an image when a predetermined time has elapsed, thereby preventing the user from missing the timing for capturing the patient's second image.

The second image may be captured with the area suspected of being a bedsore completely filled. More specifically, the second image may be captured using the selfie camera of the user terminal included in the bedsore management device 220. The bedsore management device 220 may output red light on the display unit. Users can place the selfie camera at a certain distance from the patient's suspected bedsore area and take pictures. Since the display unit emits red light, the second image captured by the selfie camera may be captured so that the red color stands out. The certain distance may be 5 cm to 20 cm or less. Because the display's red light must shine on the area suspected of having a bedsore, the distance between the patient's skin and the camera lens may need to be close.

In steps 1310 and 1320, the pressure ulcer management device 220 has been described as acquiring a still image, but it is not limited thereto. In steps 1310 and 1320, the bedsore management device 220 may capture a video. More specifically, between steps 1310 and 1320, the bedsore management device 220 may continuously capture images. The bedsore management device 220 can select at least two still images included in the video. The time between at least two still images may be a predetermined time. For example, the pressure ulcer management device 220 may acquire a first image from a video and then acquire a still image after a predetermined time as a second image.

The pressure ulcer management device 220 may acquire more images in addition to the first image and the second image. For example, the bedsore management device 220 may acquire a fourth image, which is a still image, from a video. For example, the fourth image may be an image between the first image and the second image. Additionally, the bedsore management device 220 may acquire a fifth image from the video after a predetermined time from the fourth image. For example, the fifth image may be an image after the second image. The fourth image and the fifth image may be processed in the same way as the first image and the second image and used to generate a signal indicating the possibility of a pressure ulcer in the patient. In this way, by using images other than the first image and the second image, the accuracy of the bedsore management device 220 can be further improved.

After a predetermined time has elapsed after capturing the second image, the pressure ulcer management device 220 may perform step 1330 of capturing a third image on the opposite side of the patient's area. After step 1320, the pressure ulcer management device 220 may output a message indicating to turn the patient over. The user can turn the patient over and return to a lying position. The bedsore management device 220 may output a message indicating to photograph the skin on the opposite side of the area of the patient's body where a bedsore is suspected after waiting for a predetermined time. The pressure ulcer management device 220 outputs a message indicating to capture an image when a predetermined time has elapsed, thereby preventing the user from missing the timing for capturing the patient's third image.

The patient's body part shown in the third image may be the back side of the body part in the first and second images. For example, if the first and second images are images of the patient's back, the third image may be an image of the patient's chest or stomach. The pressure ulcer management device 220 may output a message indicating that the user wishes to photograph a hairless part of the patient's body. This is because if hair is included in the third image, it may affect accuracy.

The bedsore management device 220 may output a message indicating that the distance between the photographing target and the lens will be kept approximately constant in the first, second, and third images. Additionally, when shooting the first image, second image, and third image, the user may try to keep the distance between the shooting object and the lens approximately constant.

The reason for waiting a predetermined time when shooting the third image is because the part of the patient's body to be captured in the third image was pressed during the process of shooting the first and second images. Therefore, by waiting a predetermined time, the area to be photographed in the third image can return to its original color.

The third image may be captured by completely filling the area of the patient's skin. More specifically, the third image may be captured using the selfie camera of the user terminal included in the bedsore management device 220. The bedsore management device 220 may output red light on the display unit. Users can place the selfie camera at a certain distance from the patient's suspected bedsore area and take pictures. Since the display unit emits red light, the third image captured by the selfie camera may be captured so that the red color stands out. The certain distance may be 5 cm to 20 cm or less. Because the display's red light must shine on the area suspected of having a bedsore, the distance between the patient's skin and the camera lens may need to be close.

The first to third images may need to be filled with the patient's skin. It may not matter that the first to third images are out of focus because the object and the lens are too close. If it is determined that there is a part other than skin in the first to third images based on an object recognition algorithm for recognizing skin, the pressure ulcer management device 100 may request re-photography. The user may perform steps 1310 to 1320 again at the request of the pressure ulcer management device 220.

The bedsore management device 220 may perform step 1340 of obtaining an average pixel value by averaging the values of a plurality of pixels included in the third image. The third image may include an R (red) component, a G (green) component, and a B (blue) component. The bedsore management device 220 may obtain an average pixel value by averaging pixel values for each component. For example, if the third image has a size of 1080 x 1920, the values of all 2,073,600 pixels can be averaged.

The pressure ulcer management device 220 may perform a step 1350 of extracting the average R (red) component value, the average G (green) component value, and the average B (blue) component value included in the average pixel value. The average pixel value may include an average R (red) component value, an average G (green) component value, and an average B (blue) component value.

The pressure ulcer management device 220 may perform step 1360 of determining an average GB component value based on the average G component value and the average B component value. The pressure ulcer management device 220 may perform a step of determining the sum of the average G component value and the average B component value as the average GB component value. Additionally, the pressure ulcer management device 220 may perform a step of determining the average of the average G component value and the average B component value as the average GB component value. Since the average GB component value is used later, the bedsore management device 220 may store the average GB component value.

Figure 14 is a flowchart showing the operation of a pressure ulcer management device according to an embodiment of the present disclosure. Figure 15 is a diagram for explaining the operation of a pressure ulcer management device according to an embodiment of the present disclosure.

14 and 15, the pressure ulcer management device 220 determines the first GB component value based on the first G component value and the first B component value of the pixel included in the first image 1510 ( 1410) can be performed. The first image 1510 may include a first R component value, a first G component value, and a first B component value. The bedsore management device 220 may determine the sum of the first G component value and the first B component value as the first GB component value. Additionally, the bedsore management device 220 may determine the average of the first G component value and the first B component value as the first GB component value.

The first GB component value may have a different value for each pixel included in the first image 1510. The bedsore management device 220 may determine the first GB component value by averaging the first G component value and the first B component value for each pixel included in the first image. The first GB component value may be obtained based on the first G component value and the first B component value of the corresponding pixel.

However, it is not limited to this, and the first GB component value may be one value for the entire first image 1510. For example, the bedsore management device 220 may perform a step of obtaining a first average B component value, which is the average of the B components of a plurality of pixels included in the first image. The bedsore management device 220 may obtain a first average B component value by averaging all B component values included in the first image 1510. For example, if the first image 1510 has a size of 1080 Additionally, the pressure ulcer management device 220 may perform a step of obtaining a first average G component value, which is the average of the G components of a plurality of pixels included in the first image 1510. The pressure ulcer management device 220 may obtain a first average G component value by averaging all G component values included in the first image 1510. For example, if the first image 1510 has a size of 1080 The pressure ulcer management device 220 may obtain one first GB component value for the first image 1510 based on the first average B component value and the first average G component value. The bedsore management device 220 may determine the first GB component value by averaging the first average G component value and the first average B component value. Alternatively, the pressure ulcer management device 220 may perform the step of determining the first GB component value by adding the first average G component value and the first average B component value.

The pressure ulcer management device 220 subtracts (first GB component value/average GB component value) * average R component value from the first R component value 1520 of the pixel included in the first image 1510 to perform the first subtraction. Step 1420 of obtaining a value 1530 may be performed. The pressure ulcer management device 220 may obtain the first R component value 1520 by extracting only the R component from the first image 1510. The first R component value 1520 may refer to a pixel value included in an image from which only the R component of the first image 1510 is extracted.

The pressure ulcer management device 220 may subtract the first GB component value/average GB component value)*average R component value for each pixel of the R component image of the first image 1510. The first GB component value may have a different value for each pixel of the first image 1510. The first GB component value may be obtained based on the first G component value and the first B component value of the corresponding pixel. However, it is not limited to this, and the first GB component value may be one value for the entire image. Since this has already been explained, redundant explanation will be omitted. The average GB component value may be a value obtained in step 1360. The average R component value may be obtained at step 1350.

If the first subtraction value (1530) obtained by subtracting the first R component value (1520) from the first GB component value/average GB component value) * average R component value is a negative number, the pressure ulcer management device 220 determines the first subtraction value. The value (1530) can be set to 0. In other words, the pressure ulcer management device 220 can ensure that the first deduction value 1530 has a value greater than 0. Through this, it is possible to prevent the first subtraction value 1530 from using excessive memory.

Since the first subtraction value 1530 is obtained for all pixels of the first image 1510, a set of first subtraction values 1530 may form an image. Through step 1420, the area where the user's skin has turned red due to a bedsore can be identified, regardless of the patient's skin color. The bedsore management device 100 may perform the following process to obtain the size of the reddened area.

The pressure ulcer management device 220 performs a step 1430 of obtaining a first number, which is the number of pixels whose first subtraction value 1530 is greater than or equal to a predetermined threshold value, among a plurality of pixels included in the first image 1510. It can be done. A predetermined critical component value may be predetermined. The critical component value may be a value for removing noise. The critical component value may have a value of 1/100 or less of the maximum value that a pixel can have. The critical component value may be a predetermined value, but may also be determined by the first subtraction value 1530. For example, the bedsore management device 220 may determine the average of the first subtraction values 1530 for all pixels of the first image 1510 as the critical component value.

The pressure ulcer management device 220 may determine a range of possible pressure ulcers in the first image 1510 based on the first number. However, redness on the patient's skin may be a temporary phenomenon caused by pressure. Therefore, the first number needs to be compared with the second number based on the second image 1560. For this purpose, the following process can be further performed.

The pressure ulcer management device 220 may perform step 1440 of determining a second GB component value based on the second G component value and the second B component value of the pixel included in the second image 1560. The second image 1560 may include a second R component value, a second G component value, and a second B component value. The bedsore management device 220 may determine the sum of the second G component value and the second B component value as the second GB component value. Additionally, the bedsore management device 220 may determine the average of the second G component value and the second B component value as the second GB component value.

The second GB component value may have a different value for each pixel of the second image 1560. The pressure ulcer management device 220 determines the second GB component value by averaging the second G component value and the second B component value for each pixel included in the second image. It can be done. The second GB component value may be obtained based on the second G component value and the second B component value of the corresponding pixel.

However, it is not limited to this, and the second GB component value may be one value for the entire second image 1560. For example, the bedsore management device 220 may perform a step of obtaining a second average B component value, which is the average of the B components of a plurality of pixels included in the second image. The bedsore management device 220 may obtain a second average B component value by averaging all B component values included in the second image 1550. For example, if the second image 1560 has a size of 1080 Additionally, the pressure ulcer management device 220 may perform a step of obtaining a second average G component value, which is the average of the G components of a plurality of pixels included in the second image. The bedsore management device 220 may obtain a second average G component value by averaging all G component values included in the second image 1560. For example, if the second image 1560 has a size of 1080 The pressure ulcer management device 220 may obtain one second GB component value for the second image 1560 based on the second average B component value and the second average G component value. The pressure ulcer management device 220 may determine the second GB component value by averaging the second average G component value and the second average B component value. Alternatively, the pressure ulcer management device 220 may perform the step of determining the second GB component value by adding the second average G component value and the second average B component value.

The method for obtaining the average GB component value, the first GB component value, and the second GB component value may be the same. For example, if the average GB component value is determined by the sum of the average G component value and the average B component value, the sum of the first G component value and the first B component value is determined as the first GB component value, and the second G component value is determined. The sum of the value and the second B component value can be determined as the second GB component value. However, it is not limited to this.

The pressure ulcer management device 220 obtains a second subtracted value by subtracting (second GB component value/average GB component value) * average R component value from the second R component value of the pixel included in the second image 1560. Step 1450 may be performed. The bedsore management device 220 may obtain a second R component value 1570 by extracting only the R component from the second image 1560. The second R component value 1570 may refer to a pixel value included in an image from which only the R component of the second image 1560 is extracted.

The pressure ulcer management device 220 may subtract the second GB component value/average GB component value)*average R component value for every pixel of the R component image of the second image 1560. The second GB component value may have a different value for each pixel of the second image 1560. The second GB component value may be obtained based on the second G component value and the second B component value of the corresponding pixel. However, it is not limited to this, and the second GB component value may be one value for the entire image. Since this has already been explained, redundant explanation will be omitted. The average GB component value may be a value obtained in step 1360. The average R component value may be obtained at step 1350.

If the second subtraction value (1580) obtained by subtracting the second R component value (1570) from the second GB component value/average GB component value) * average R component value is a negative number, the pressure ulcer management device 220 determines the second subtraction value The value (1580) can be set to 0. In other words, the pressure ulcer management device 220 can ensure that the second deduction value 1580 has a value greater than 0. Through this, it is possible to prevent the second subtraction value 1580 from using excessive memory.

Since the second subtraction value 1580 is obtained for all pixels of the second image 1560, a set of the second subtraction values 1580 can form an image. Through step 1450, the area where the user's skin has turned red due to bedsores can be confirmed regardless of the patient's skin color. Additionally, the first subtraction value 1530 obtained in step 1420 may be compared with the second subtraction value 1580 obtained in step 1450. The pressure ulcer management device 100 may perform the following process to obtain the size of the reddened area and compare the first subtraction value 1530 and the second subtraction value 1580.

The pressure ulcer management device 220 performs a step 1460 of obtaining a second number, which is the number of pixels whose second subtraction value 1580 is greater than or equal to a predetermined threshold value, among a plurality of pixels included in the second image 1560. It can be done. A predetermined critical component value may be predetermined. The critical component value may be a value for removing noise. The critical component value may be a predetermined value, but may also be determined by the first subtraction value 1530. For example, the bedsore management device 220 may determine the average of the first subtraction values 1530 for all pixels of the first image 1510 as the critical component value. The pressure ulcer management device 220 may set the standard for obtaining the second number to be the same as the standard for obtaining the first number.

The pressure ulcer management device 220 may determine a range of possible pressure ulcers in the first image 1510 based on the first number. However, redness on the patient's skin may be a temporary phenomenon caused by pressure. Therefore, the first number needs to be compared with the second number based on the second image 1560. For this purpose, the following process can be further performed.

If the value obtained by subtracting the second number from the first number is less than or equal to a predetermined threshold number, the pressure ulcer management device 220 may perform step 1470 of generating a signal indicating that there is a possibility of a pressure ulcer in the patient. This is because it means that the skin color does not return from red to normal. The user terminal included in the bedsore management device 220 may output a signal indicating that there is a possibility of a bedsore in the patient. Therefore, the user can take necessary action for the patient. For example, you can visit a medical institution or ask for help from a professional medical professional. Additionally, the user terminal included in the bedsore management device 220 can transmit a signal indicating the possibility of a bedsore in the patient to a remote medical server. Additionally, the user terminal may transmit at least one of the first image 1510 or the second image 1560 to the remote medical server. A remote medical server may be included in the bedsore management device 220. The remote medical server can use the first image 1510 or the second image 1560 as the current patient's pressure ulcer image data. The remote medical server included in the bedsore management device 220 may apply a machine learning model to the current patient's bedsore image data to determine information about the current patient's bedsores and/or pressure ulcer treatment information (S350). Additionally, the remote medical server included in the bedsore management device 220 may transmit the determined information about the current patient's bedsores and/or pressure ulcer treatment information to the user terminal. The user can remotely check information about the current patient's bedsores and/or pressure ulcer treatment information. Through the process shown in Figure 3, more detailed information about bedsores can be provided to the user than through the process shown in Figures 13 and 14. Through the processes of FIGS. 13 and 14, the same process as that of FIG. 3 can be performed on data of a patient suspected of having bedsores. Machine learning models require a lot of processing resources. Therefore, if many users wish to obtain information about the current patient's bedsores and/or bedsore treatment information through the process shown in FIG. 3, the computing performance of the external medical server may need to be excessively high. However, by performing the processes of FIGS. 13 and 14 before the process of FIG. 3, the computing processing capacity of the external medical server may be somewhat low. The processes of Figures 13 and 14 can be implemented with relatively low processing capabilities, so there is no need for an external medical server to process them directly, and only the data required by the processes of Figures 13 and 14 can be processed by an external medical server. Because.

Additionally, the pressure ulcer management device 220 may perform step 1470 of generating a signal indicating that there is a possibility of pressure ulcers in the patient when the value obtained by subtracting the second number from the first number is greater than or equal to a predetermined threshold number.

Figure 16 is a flowchart showing the operation of a pressure ulcer management device according to an embodiment of the present disclosure.

The pressure ulcer management device 220 may further perform the following process in order to perform the step 1470 of generating a signal indicating that the patient in FIG. 14 has a possibility of having a pressure ulcer.

According to an embodiment of the present disclosure, the bedsore management device 220 may acquire one first GB component value for the first image. The pressure ulcer management device 220 may perform a step 1610 of obtaining a first difference value obtained by subtracting the average GB component value from the first GB component value and then taking the absolute value. Additionally, according to an embodiment of the present disclosure, the bedsore management device 220 may acquire one first GB component value for the first image. The pressure ulcer management device 220 may perform a step 1620 of obtaining a second difference value obtained by subtracting the average GB component value from the second GB component value and then taking the absolute value.

The pressure ulcer management device 220 provides pressure ulcers to the patient when the value obtained by subtracting the second number from the first number is less than or equal to a predetermined threshold number and the first or second difference value is less than or equal to a predetermined critical difference value. A step 1630 of generating a signal indicating that there is a possibility may be performed.

In the case of severe bedsores, the user can immediately check with the eyes, so the user can know whether the patient has bedsores without the process shown in Figure 13 or Figure 14. At this time, it may be more important for the user to decide what treatment to give to the patient using the method shown in FIG. 3. The pressure ulcer management device 220 can simply confirm the possibility of an initial pressure ulcer based on the similarity of the colors of the images through the process of FIGS. 13 and 14. The first to third images may be images taken in similar situations. Therefore, it is possible to simply check whether the first to third images are similar by checking whether the first difference value or the second difference value is less than or equal to a predetermined threshold difference value. When the pressure ulcer management device 220 generates a signal indicating the possibility of a pressure ulcer in the patient in step 1630, the steps of FIG. 3 are performed by transmitting at least one of the first image or the second image to an external medical server. It can be done. In addition, if the pressure ulcer management device 220 does not generate a signal indicating the possibility of a pressure ulcer in the patient through step 1630, it outputs a message suggesting that the user perform the steps of FIGS. 13 and 14 again, or You can print a message providing contact information for non-face-to-face contact with a medical professional. Users can receive consultation from a medical professional non-face-to-face.

According to various embodiments of the present disclosure, the bedsore management device 220 may obtain the first GB component value for each pixel of the first image. The bedsore management device 220 may obtain a first average GB component value by averaging the first GB component values of all pixels of the first image. The pressure ulcer management device 220 may perform a step 1610 of obtaining a first difference value obtained by subtracting the average GB component value from the first average GB component value and then taking the absolute value. Additionally, the pressure ulcer management device 220 may obtain the second GB component value for each pixel of the second image. The bedsore management device 220 may obtain a second average GB component value by averaging the second GB component values of all pixels of the second image. The pressure ulcer management device 220 may perform a step 1620 of obtaining a second difference value obtained by subtracting the average GB component value from the second average GB component value and then taking the absolute value. The pressure ulcer management device 220 provides pressure ulcers to the patient when the value obtained by subtracting the second number from the first number is less than or equal to a predetermined threshold number and the first or second difference value is less than or equal to a predetermined critical difference value. A step 1630 of generating a signal indicating that there is a possibility may be performed.

The pressure ulcer management device 220 provides pressure ulcers to the patient when the value obtained by subtracting the second number from the first number is greater than or equal to a predetermined threshold number and the first or second difference value is less than or equal to the predetermined critical difference value. A step 1630 of generating a signal indicating that there is a possibility may be performed. The pressure ulcer management device 220 provides pressure ulcers to the patient when the value obtained by subtracting the second number from the first number is more than a predetermined threshold number and the first difference value or the second difference value is more than a predetermined threshold difference value. A step 1630 of generating a signal indicating that there is a possibility may be performed. The pressure ulcer management device 220 provides pressure ulcers to the patient when the value obtained by subtracting the second number from the first number is less than or equal to a predetermined threshold number and the first or second difference value is greater than or equal to the predetermined threshold difference value. A step 1630 of generating a signal indicating that there is a possibility may be performed.

The pressure ulcer management device 220 may further perform the following process in order to perform the step 1470 of generating a signal indicating that the patient in FIG. 14 has a possibility of having a pressure ulcer.

If the first number is more than 1/2 of the total number of pixels in the first image, and the value obtained by subtracting the second number from the first number is less than or equal to a predetermined threshold number, a signal is generated indicating that there is a possibility of pressure ulcers in the patient. You can follow the steps: That is, the bedsore management device 220 may further determine whether the first number is more than 1/2 of the total number of pixels of the first image. As already described, the first image and the second image may be captured with the area suspected to be a bedsore completely filled. Since the first number is the range suspected of being bedsores, the bedsore management device 220 performs a step of generating a signal indicating that there is a possibility of bedsores in the patient only when the range suspected of being bedsores is at least 1/2 or more, thereby managing bedsores. Device 220 can increase reliability.

If the first number is less than 1/2 of the total number of pixels in the first image, and the value obtained by subtracting the second number from the first number is less than or equal to a predetermined threshold number, a signal is generated indicating that there is a possibility of pressure ulcers in the patient. You can follow the steps: If the first number is less than 1/2 of the total number of pixels in the first image, and the value obtained by subtracting the second number from the first number is more than a predetermined threshold number, a signal is generated indicating that there is a possibility of pressure ulcers in the patient. You can follow the steps: If the first number is more than 1/2 of the total number of pixels in the first image, and the value obtained by subtracting the second number from the first number is more than a predetermined threshold number, a signal is generated indicating that there is a possibility of pressure ulcers in the patient. You can follow the steps:

So far, we have looked at various embodiments. A person skilled in the art to which this disclosure pertains will understand that this disclosure may be implemented in a modified form without departing from the essential characteristics of the present disclosure. Therefore, the disclosed embodiments should be considered from an illustrative rather than a restrictive perspective. The scope of the present disclosure is indicated in the claims rather than the foregoing description, and all differences within the equivalent scope should be construed as being included in the present disclosure.

The apparatus for managing bedsores and the operating method of the apparatus for managing bedsores of the present disclosure enable medical services to be provided by machine learning the relationship between pictures of bedsores and information about bedsores to evaluate the stage of bedsores and guide dressing treatments accordingly. It is expected that it will contribute to increasing efficiency. In addition, the difference in nursing performance depending on the medical professional's experience can be reduced and the workload can be expected to be reduced. Furthermore, changes in the patient's pressure ulcer status can be continuously checked through system linkage without using special equipment, allowing intensive management of pressure ulcers. This can reduce the mortality rate of patients with pressure ulcers, shorten the length of hospital stay, and reduce medical expenses. You will be able to do it.

Meanwhile, the above-described embodiments of the present disclosure can be written as a program that can be executed on a computer, and can be implemented in a general-purpose digital computer that operates the program using a computer-readable recording medium. The computer-readable recording media includes storage media such as magnetic storage media (eg, ROM, floppy disk, hard disk, etc.) and optical read media (eg, CD-ROM, DVD, etc.).

Claims (5)

In a method of operating a bedsore management device, Capturing, by the bedsore management device, a first image of an area suspected of having a bedsore in a patient; A predetermined time after capturing the first image, the bedsore management device captures a second image of the area where the patient's bedsore is suspected; capturing, by the pressure ulcer management device, a third image of the opposite side of the area of the patient after the predetermined time after capturing the second image; obtaining an average pixel value by averaging the values of a plurality of pixels included in the third image; extracting the average R (red) component value, the average G (green) component value, and the average B (blue) component value included in the average pixel value; determining an average GB component value based on the average G component value and the average B component value; determining a first GB component value based on a first G component value and a first B component value of a pixel included in the first image; Obtaining a first subtraction value by subtracting (the first GB component value/the average GB component value) * the average R component value from the first R component value of a pixel included in the first image; Obtaining a first number, which is the number of pixels among a plurality of pixels included in a first image, where the first subtraction value is greater than or equal to a predetermined threshold value; determining a second GB component value based on a second G component value and a second B component value of a pixel included in the second image; Obtaining a second subtraction value by subtracting (the second GB component value/the average GB component value) * the average R component value from the second R component value of the pixel included in the second image; Obtaining a second number, which is the number of pixels among a plurality of pixels included in a second image for which the second subtraction value is greater than or equal to the predetermined threshold value; and A method of operating a pressure ulcer management device comprising generating a signal indicating that there is a possibility of a pressure ulcer in a patient when a value obtained by subtracting the second number from the first number is less than or equal to a predetermined threshold number. According to claim 1, Determining the average GB component value includes determining the average GB component value by averaging the average G component value and the average B component value, Determining the first GB component value includes determining the first GB component value by averaging the first G component value and the first B component value for each pixel included in the first image; , Determining the second GB component value includes determining the second GB component value by averaging the second G component value and the second B component value for each pixel included in the second image. How the pressure ulcer management device works. According to claim 1, The step of determining the first GB component value is, Obtaining a first average G component value that is the average of G components of a plurality of pixels included in the first image; Obtaining a first average B component value that is the average of B components of a plurality of pixels included in the first image; and A step of determining the first GB component value by averaging the first average G component value and the first average B component value, The step of determining the second GB component value is, Obtaining a second average G component value that is the average of G components of a plurality of pixels included in the second image; Obtaining a second average B component value that is the average of B components of a plurality of pixels included in the second image; and A method of operating a pressure ulcer management device comprising determining the second GB component value by averaging the second average G component value and the second average B component value. According to claim 3, Generating a signal indicating the possibility of a pressure ulcer in the patient comprises: Obtaining a first difference value obtained by subtracting the average GB component value from the first GB component value and taking the absolute value; Obtaining a second difference value obtained by subtracting the average GB component value from the second GB component value and taking an absolute value; and When the value obtained by subtracting the second number from the first number is less than or equal to a predetermined threshold number and the first difference value or the second difference value is less than or equal to a predetermined critical difference value, there is a possibility of bedsores in the patient. A method of operating a pressure ulcer management device comprising generating an indicative signal. According to claim 1, Generating a signal indicating the possibility of a pressure ulcer in the patient comprises: If the first number is more than 1/2 of the total number of pixels of the first image, and the value obtained by subtracting the second number from the first number is less than or equal to a predetermined threshold number, this indicates that there is a possibility of bedsores in the patient. A method of operating a pressure ulcer management device comprising generating a signal.

Images