TWI861690B - Sepsis Care Feedback System - Google Patents

Abstract

The invention comprises a database, a display module and a human-computer interaction module electrically connected to a processing module; the database stores a plurality of patient data and a sepsis detection model; each of the patient data comprises a physiological information, and the sepsis detection model is an artificial intelligence model; when the processing module determines through the sepsis detection model that the physiological information of one of the patient data meets the sepsis detection criteria, Under normal circumstances, the processing module calculates a prediction probability through the sepsis detection model, and displays a possible diagnosis message and a confirmation screen through the display module, and the display module displays the prediction probability and a sepsis reference drug order message; the present invention can provide a reference sepsis risk assessment and the sepsis reference drug order message, so as to facilitate the emergency room to treat sepsis patients more efficiently.

Description

Sepsis Care Feedback System

A medical care feedback system, especially a sepsis care feedback system.

Every second counts in the emergency room. Every patient sent to the hospital emergency room may be admitted to the Intensive Care Unit (ICU) for treatment due to varying degrees of pain. However, when the medical team is treating the patient's more obvious injuries, the possibility of the patient getting sepsis is easily overlooked.

Patients with sepsis usually do not show any obvious symptoms at first, so doctors may not realize that they should be tested for sepsis until later, which leads to later detection of sepsis and later provision of sepsis-related medical care. However, sepsis is a disease that is better treated earlier. When relevant treatment is delayed, it is difficult for patients with sepsis to improve their condition.

Currently, doctors in the ICU can only rely on their diagnostic experience to determine whether a patient needs to undergo sepsis-related tests. However, doctors are also human beings, and humans are more likely to have subjective judgments. When doctors in the ICU rely solely on experience to determine whether each patient needs to be tested for sepsis, subjective thinking is more likely to lack a unified evaluation standard and affect the results of this decision. In the worst case, doctors may misjudge that patients do not need to be tested for sepsis, which may worsen the condition of patients who actually have sepsis.

Furthermore, when there are many patients in the ICU of a hospital, doctors and medical staff need to efficiently diagnose whether each patient has the possibility of sepsis. This process requires a huge amount of human resources and time. With limited time and energy, doctors and medical staff may also make mistakes in diagnosing sepsis. This will result in sepsis patients not being able to receive proper treatment early, and may also lead to worsening of their condition.

In view of the above problems, the present invention provides a sepsis care feedback system that can automatically determine whether a patient is more likely to have sepsis based on the patient's physiological information, and generate and present this information for the medical team in the emergency room to know. The medical team in the emergency room can more efficiently decide and diagnose whether the patient is indeed at high risk of sepsis based on this information with reference value.

The sepsis care feedback system includes: A database storing a plurality of patient data, a sepsis detection model, a plurality of sepsis classification options and a sepsis medication table; wherein each of the patient data includes a physiological information; wherein the sepsis detection model is an artificial intelligence model; A processing module electrically connected to the database, and judging whether the physiological information of each of the patient data meets a sepsis detection standard through the sepsis detection model; A display module electrically connected to the processing module; A human-computer interaction module electrically connected to the processing module; Wherein, when the processing module determines that the physiological information of one of the patient data meets the sepsis detection standard through the sepsis detection model, the processing module calculates a prediction probability through the sepsis detection model, and displays a possible diagnosis message and a confirmation screen through the display module; Wherein, when the processing module receives a sepsis confirmation signal from the human-computer interaction module, the processing module displays the prediction probability and the sepsis classification options through the display module; When the processing module receives a sepsis type data from the human-computer interaction module, the processing module generates a sepsis reference drug list message according to the sepsis type data and compares the sepsis medication list, and displays the sepsis reference drug list message through the display module.

In this way, the medical team can see the prompt of the present invention indicating that the patient may have sepsis from the display module and make appropriate medical treatment. When the medical team confirms through the human-computer interaction module that the patient does have sepsis, the medical team can further view the sepsis classification options from the display module and operate the human-computer interaction module to input the patient's sepsis type, so as to generate the sepsis reference medication order information according to the medication logic set in the sepsis medication table. Therefore, the present invention can not only assist the medical team in the emergency room to more efficiently determine whether a patient is more likely to suffer from sepsis, but also more efficiently provide a reference medicine list for the medical team to refer to, thereby increasing the speed at which the medical team determines the type of medicine to be used for sepsis patients.

Please refer to FIG. 1 , the present invention is a sepsis care feedback system. The sepsis care feedback system includes a database 10, a processing module 20, a display module 30 and a human-machine interaction module 40. The processing module 20 is electrically connected to the database 10, the display module 30 and the human-machine interaction module 40 respectively.

The database 10 stores a plurality of patient data, a sepsis detection model, a plurality of sepsis classification options, and a sepsis medication table. Each of the patient data includes physiological information. The sepsis detection model is an artificial intelligence (AI) model, and the processing module 20 determines whether the physiological information of each of the patient data meets a sepsis detection standard through the sepsis detection model.

In one embodiment of the present invention, the sepsis detection model is a mature AI model that can help accurately calculate the risk of a patient suffering from sepsis to provide reference information for medical staff. Before accurate calculation, the processing module 20 first searches, collects, filters and integrates the information of one of the patient data in the database 10 through the sepsis detection model, so that the sepsis detection model can obtain all the information required for calculation. Then, after the sepsis detection model is calculated, the processing module 20 uses the sepsis detection model to determine whether the physiological information of one of the patient data meets the sepsis detection standard. The sepsis detection standard can be understood as a threshold range of multiple patient physiological values. Furthermore, the sepsis detection model uses text recognition, phrase recognition, and code recognition to search and collect information related to one of the patient data from the database 10, then filters out information not related to sepsis, and then integrates the remaining information related to sepsis. The text recognition, phrase recognition, and code recognition methods used by the sepsis detection model can be recognition technologies used in the existing AI field, so they are not elaborated here.

When the processing module 20 determines that the physiological information of one of the patient data meets the sepsis detection standard through the sepsis detection model, the processing module 20 calculates a prediction probability through the sepsis detection model and generates a possible diagnosis message. On the contrary, when the processing module 20 determines that the physiological information of one of the patient data does not meet the sepsis detection standard through the sepsis detection model, the processing module 20 does not calculate the prediction probability through the sepsis detection model and generate the possible diagnosis message. The calculation of the prediction probability is the probability of the processing module 20 predicting the diagnosis of sepsis using the sepsis detection model. This probability will change due to different physiological values of the patient.

Please refer to FIG. 2 , the processing module 20 further generates an overview screen 100, and the processing module 20 displays the overview screen 100 through the display module 30. The overview screen 100 displays the patient data stored in the database 10, that is, the basic data of all patients in the computer system of an intensive care unit (ICU) of a hospital. As shown in FIG. 2 , in this embodiment, the patient data includes information such as name, age, gender, bed data, number of emergency examinations, number of X-rays, etc. In addition, the processing module 20 also displays the possible diagnosis information 101 corresponding to the corresponding part of the patient data in the overview screen 100 through the display module 30. For example, according to the example of FIG. 2 , Mr. Wang and Ms. Wang in the ICU are considered to be likely to have sepsis through the judgment of the sepsis detection model by the processing module 20, while the other patients in the ICU are not at risk of sepsis. The possible diagnosis information 101 can be provided to a user, that is, the medical team in the ICU, to make a selection by operating the human-computer interaction module 40.

In this embodiment, the processing module 20 is a processor, the display module 30 is a screen, and the human-machine interaction module 40 is a mouse and a keyboard. In other embodiments, the processing module 20 is a network server, and the display module 30 and the human-machine interaction module 40 are a touch panel of a smart mobile device. The smart mobile device is, for example, a tablet computer or a smart phone, and the smart mobile device can be connected to the network server via a wireless network.

Referring to FIG. 3 , when the possible diagnosis message 101 is selected, the processing module 20 further generates a confirmation screen 110, and the processing module 20 displays the confirmation screen 110 through the display module 30. The confirmation screen 110 includes a non-septicemia option 111, a possible sepsis option 112, and a confirmed sepsis option 113. When the user selects the non-septicemia option 111 or the possible sepsis option 112 by operating the human-computer interaction module 40, it means that the user has manually confirmed that the patient does not have the risk of having sepsis. In this way, the human-machine interaction module 40 generates a non-septicemia signal to the processing module 20, so that the processing module 20 jumps back to display the overview screen 100 through the display module 30 and cancels the display of the selected possible diagnosis message 101. When the user selects the confirmed sepsis option 113 by operating the human-machine interaction module 40, it means that the user has manually further confirmed that the patient has reached the risk of having sepsis. In this way, the human-machine interaction module 40 generates a confirmed sepsis signal to the processing module 20. When the processing module 20 receives the confirmed sepsis signal from the human-machine interaction module 40, the processing module 20 displays the predicted probability 114 through the display module 30.

Please refer to FIG. 4 , the processing module 20 further displays a classification screen 120 through the display module 30, and the classification screen 120 displays the sepsis classification options for selection. The sepsis classification options are established based on part of the content of severe sepsis patients (ABC Model; ABC). Among them, A of ABC represents eight major infection sites, B represents failure of seven major organs, and C represents severe sepsis. The sepsis classification options displayed for selection on the classification screen 120 include eight major infection site types and a regional type of sepsis. The eight infection site categories include a central nervous system infection option 121, a respiratory system infection option 122, a urinary system infection option 123, a hepatobiliary gastrointestinal system infection option 124, a bone and joint system infection option 125, a soft tissue system infection option 126, a cardiovascular system infection option 127, and a primary unknown system infection option 128. A sepsis-stricken area category includes a community option 129A and a medical care center option 129B.

In the example of FIG. 4 , the user selects the hepatobiliary gastrointestinal infection option 124 and the medical care center option 129B through the human-computer interaction module 40, and the human-computer interaction module 40 generates sepsis type data corresponding to hepatobiliary gastrointestinal infection and infection in a medical care center, and so on.

As shown in FIG. 5 , when the processing module 20 receives the sepsis type data from the human-machine interaction module 40, the processing module 20 generates a sepsis reference medication order message 130 by comparing the sepsis type data with the sepsis medication table, and displays the sepsis reference medication order message 130 through the display module 30. The sepsis reference medication order message 130 includes sepsis recommended medication information 131. The relationship between the sepsis type data and the sepsis recommended medication information 131 is established by the sepsis medication table, and the comparison relationship established by the sepsis medication table is shown in Table 1 below: Community Medical Care Center Central nervous system infection Ceftriaxone Vancomycin Cefepime Vancomycin Respiratory system infection Ceftriaxone Levofloxacin Cefepime Levofloxacin Vancomycin Urinary tract infection Ceftriaxone Cefepime Infection of the liver, gallbladder, and gastrointestinal system Ertapenem Meropenem Infection of the bone and joint system Oxacillin Vancomycin Software infection Ceftriaxone Ciprofloxacin Clindamycin Vancomycin Cefepime Cardiovascular system infection Oxacillin Gentamicin (GM) Vancomycin Cefepime Unknown origin system infected Ceftriaxone Cefepime Vancomycin Table 1

According to Table 1, the aforementioned example of the sepsis type data of hepatobiliary gastrointestinal system infection and infection in a medical care center is that the processing module 20 compares the data with Table 1 and recommends the use of ceftriaxone as the sepsis recommended medication information 131. In addition, according to the sepsis type data, the processing module 20 can also display a bit of information 132 and a booster injection drug information 133 that the patient needs to use in the sepsis reference medication message 130 through the display module 30. The drip information 132 and the booster injection drug information 133 are options that can be disclosed. When clicked and expanded, the information details of the drip information 132 and the booster injection drug information 133 can be presented in list form.

As shown in FIG1 , in the present embodiment, the processing module 20 further includes an emergency and critical care testing unit 21, an abnormal blood draw tracking reminder unit 22, a sepsis antibiotic automatic guidance unit 23, a hemodynamic monitor evaluation unit 24, an infusion reminder unit 25, a medication reminder unit 26 and a personalized feedback monitoring unit 27. Among them, the emergency and critical illness testing unit 21, the abnormal blood drawing tracking reminder unit 22, the sepsis antibiotic automatic guidance unit 23, the hemodynamic monitor evaluation unit 24, the infusion reminder unit 25, the medication reminder unit 26 and the personalized feedback monitoring unit 27 are respectively electrically connected to the database 10 and the display module 20, and the infusion reminder unit 25, the medication reminder unit 26 and the personalized feedback monitoring unit 27 are also respectively electrically connected to the human-computer interaction module 40.

The database 10 further stores a present time, a first time period, a second time period, a care time, an infusion time threshold, a medication time threshold, a plurality of care note messages, a plurality of care attention messages, a lactate value threshold, a cardiac output threshold and a water acid value threshold.

Please refer to FIG6 . When the processing module 20 receives the confirmed sepsis signal, the emergency and critical care test unit 21 generates a critical care test table 140 corresponding to the current time, and displays the critical care test table 140 through the display module 30. Moreover, the emergency and critical care test unit regenerates the critical care test table 140 every time the first time period passes. In the present embodiment, the first time period is a 24-hour time period, so as shown in FIG6 , the critical care test table 140 is not only generated at the current time, but also updated and generated after 24 hours, 48 hours, and 72 hours. In other words, the critical care test table 140 includes expandable options, and some of these options are divided into intervals with the first time period of every 24 hours. As shown in FIG6 , the critical examination table 140 generated at the current time is clicked and expanded, and the contents displayed after expansion include a plurality of required examination items 141 and a plurality of optional examination items 142 for selection. The user can select or deselect the optional examination items 142 through the human-machine interaction module 40. In this embodiment, the database 10 may further store other time limits to facilitate reminders at other time periods, such as the critical examination table 140 updated and generated one week from the current time.

Please refer to FIG. 7 , after the processing module 20 generates the sepsis reference prescription message 130, and each time the infusion time threshold is passed, the infusion reminder unit 25 generates an infusion reminder message 150, and displays the infusion reminder message 150 through the display module 30. Moreover, after the processing module 20 generates the sepsis reference prescription message 130, and each time the medication time threshold is passed, the medication reminder unit 26 generates a medication reminder message 160, and displays the medication reminder message 160 through the display module 30. The infusion reminder message 150 includes an infusion volume 151 and an infusion time limit 152, so that the user can know the urgency of infusion and the amount of infusion given to the patient. In other embodiments, the medication reminder message 160 may also include a dosage 161 and a medication time limit, so that the user knows the urgency and dosage of medication to be administered to the patient. For example, in the example of Fig. 7, the infusion time limit 152 is a time limit of 3 hours.

As shown in FIG8 , the display module 30 further displays a dosing time field 170. The user can fill in the actual dosing time in the dosing time field 170 by operating the human-machine interaction module 40. When the dosing reminder unit 26 receives a dosing time input into the dosing time field 170 from the human-machine interaction module 40, the dosing reminder unit 26 generates and displays the dosing reminder message 160 every time the dosing time threshold is passed starting from the dosing time.

Please refer to FIG9 , when the processing module 20 generates the sepsis reference prescription message 130, the personalized feedback monitoring unit 27 starts to count an emergency room retention time 181, and generates a combined care table 180, and displays the combined care table 180 and the emergency room retention time 181 through the display module 30. The combined care table 180 includes at least one care action to be performed at each interval of the second time period generated according to the sepsis reference prescription message 130. In this embodiment, the second time period is a one-hour period, so as shown in FIG9 , the combined care table 180 displays at least one care action to be performed every hour starting from the current time. For example, at the first hour from the current time, a reminder is given to draw blood for a blood test to detect a first lactate value, perform cell culture, prescribe antibiotics, and prescribe pressors, and at the second hour from the current time, a reminder is given to administer fluids. In this embodiment, the database 10 may further store other time limits to facilitate giving reminders at other time periods, such as at the fourth hour from the current time, a reminder is given to draw blood for a second blood test to detect a second lactate value.

With respect to the at least one care action that the user is reminded of by the combined care table 180, when the user performs the care action, the user can generate a completed care message through the human-machine interaction module 40. When the personalized feedback monitoring unit 27 receives the completed care message from the human-machine interaction module 40, the personalized feedback monitoring unit 27 generates a completed care message and displays the completed care message through the display module. The completed care message can be divided into at least one completed message 182 and at least one opened message 183. The at least one executed message 182 corresponds to an action of performing a blood test to measure lactate value or performing cell culture, and the at least one prescribed message 183 corresponds to an action of prescribing antibiotics or prescribing pressors.

Furthermore, at each interval of the second time period, the personalized feedback monitoring unit 27 counts down the care time 184. The care time 184 can remind the user of the time limit for treating the patient, so as to urge the user to treat the patient on time. In addition, each of the at least one care actions has the aforementioned infusion time limit 152 or the medication time limit.

When the personalized feedback monitoring unit 27 does not receive the executed care message from the human-machine interaction module 40 before the care time 184 countdown returns to zero, the personalized feedback monitoring unit 27 generates a timeout message 185 and displays the timeout message 185 through the display module 30. For example, in FIG9 , although the user has completed the cell culture, due to the timeout, a period of time has passed after the care time 184 countdown returns to zero, and the corresponding timeout message 185 is still displayed on the display module 30. In addition, the user has timed out and has not yet opened the booster, so the corresponding timeout message 185 is displayed on the display module 30.

In addition, when the personalized feedback monitoring unit 27 has not received the executed care message from the human-machine interaction module 40, the personalized feedback monitoring unit 27 generates a not yet care message. The not yet care message can be divided into at least one not yet executed message and at least one not yet opened message 186. In addition, the care note messages stored in the database 10 correspond to the at least one care action. When the personalized feedback monitoring unit 27 receives a signal for selecting the at least one care action from the human-machine interaction module 40, the personalized feedback monitoring unit 27 displays one of the care note messages 187 corresponding to the at least one care action through the display module 30.

Furthermore, when the at least one care action is an Nth blood drawing, and the abnormal blood drawing tracking reminder unit 22 determines that a lactate value in the blood is higher than the lactate value threshold, the abnormal blood drawing tracking reminder unit 22 notifies the personalized feedback monitoring unit 27 to display one of the care attention messages 188 corresponding to the at least one care action of an (N+1)th blood drawing through the display module 30. Wherein, N is a positive integer, and the so-called (N+1)th blood drawing means the next blood drawing after the Nth time. In the example of FIG. 9 , because the patient's lactate value during the first blood draw was higher than the lactate value threshold, the display module 30 displays one of the care notice messages 188 next to the description of the second blood draw to help the user notice the abnormal situation that the patient's lactate value was too high during the last blood draw.

In addition, the physiological information of each patient data further includes a cardiac output and a water acid value. When the hemodynamic monitor evaluation unit 24 determines that the cardiac output of one of the patient data is less than the cardiac output threshold or determines that the water acid value of one of the patient data is greater than the water acid value threshold, the hemodynamic monitor evaluation unit 24 generates a hemodynamic abnormality message and displays the hemodynamic abnormality message through the display module (not shown). In this way, the user can notice the abnormal hemodynamic condition of the patient.

The user of the present invention, that is, the medical team, can see the prompt that the patient may have sepsis from the display module 30 and make appropriate medical treatment. When the medical team confirms that the patient does have sepsis through the human-machine interaction module 40, the medical team can further view the sepsis classification options from the display module 30. The medical team can further operate the human-machine interaction module 40 to input the patient's sepsis type, so that the sepsis reference medication order message 130 is generated according to the medication logic set in the sepsis medication table. Therefore, the present invention can not only assist the medical team in the emergency room to more efficiently determine whether a patient is more likely to have sepsis, but also more efficiently provide a reference list of medicines for the medical team to refer to, thereby increasing the speed at which the medical team determines the type of medicine to be used for septic patients. The present invention can also further assist the medical team to continuously monitor the patient's biological condition and monitor whether the medical team has given the patient appropriate medical treatment in a timely manner, so as to help the patient recover from sepsis in the shortest possible time.

10: Database 20: Processing module 21: Emergency and critical care testing unit 22: Abnormal blood draw tracking reminder unit 23: Antibiotic automatic guidance unit for sepsis 24: Hemodynamic monitor evaluation unit 25: Infusion reminder unit 26: Medication reminder unit 27: Personalized feedback monitoring unit 30: Display module 40: Human-computer interaction module 100: Overview screen 101: Possible diagnosis message 110: Confirmation screen 111: Non-septicemia option 112: Possible sepsis option 113: Confirmed sepsis option 114: Prediction probability 120: Classification screen 121: Central nervous system infection options 122: Respiratory system infection options 123: Urinary system infection options 124: Hepatobiliary and gastrointestinal system infection options 125: Osteoarthritis system infection options 126: Soft tissue system infection options 127: Cardiovascular system infection options 128: Primary unknown system infection options 129A: Community options 129B: Medical care center options 130: Sepsis reference drug list information 131: Sepsis recommended medication information 132: Intravenous drip information 133: Boost injection drug information 141: Mandatory inspection items 142: Optional inspection items 151: Infusion volume 152: Infusion time limit 161: Dosage 170: Execution time field 180: Combined care table 181: Emergency room stay time 182: Executed message 183: Issued message 184: Care time 185: Timeout message 186: Not yet issued message 187: Care notes message 188: Care attention message

FIG. 1 is a system block diagram of a sepsis care feedback system of the present invention. FIG. 2 is a schematic diagram of a display module of the sepsis care feedback system of the present invention displaying an overview screen. FIG. 3 is a schematic diagram of the display module of the sepsis care feedback system of the present invention displaying a confirmation screen. FIG. 4 is a schematic diagram of the display module of the sepsis care feedback system of the present invention displaying a classification screen. FIG. 5 is a schematic diagram of the display module of the sepsis care feedback system of the present invention displaying a sepsis reference drug list message. FIG. 6 is a schematic diagram of the display module of the sepsis care feedback system of the present invention displaying a critical examination table. FIG7 is a schematic diagram of the display module of the sepsis care feedback system of the present invention displaying an infusion reminder message and a medication reminder message. FIG8 is a schematic diagram of the display module of the sepsis care feedback system of the present invention displaying a medication time field. FIG9 is a schematic diagram of the display module of the sepsis care feedback system of the present invention displaying a combined care table.

10: Database

20: Processing module

21: Emergency and critical care testing unit

22: Abnormal blood draw tracking reminder unit

23: Septicemia antibiotic automatic guidance unit

24: Hemodynamic monitor evaluation unit

25: Infusion reminder unit

26: Medication reminder unit

27: Personalized feedback monitoring unit

30: Display module

40: Human-computer interaction module

Claims (10)

A sepsis care feedback system includes: A database storing a plurality of patient data, a sepsis detection model, a plurality of sepsis classification options and a sepsis medication table; wherein each of the patient data includes physiological information; wherein the sepsis detection model is an artificial intelligence model; A processing module electrically connected to the database, and judging whether the physiological information of each of the patient data meets a sepsis detection standard through the sepsis detection model; A display module electrically connected to the processing module; A human-computer interaction module electrically connected to the processing module; Wherein, when the processing module determines that the physiological information of one of the patient data meets the sepsis detection standard through the sepsis detection model, the processing module calculates a prediction probability through the sepsis detection model, and displays a possible diagnosis message and a confirmation screen through the display module; Wherein, when the processing module receives a sepsis confirmation signal from the human-computer interaction module, the processing module displays the prediction probability and the sepsis classification options through the display module; When the processing module receives a sepsis type data from the human-computer interaction module, the processing module generates a sepsis reference drug list message according to the sepsis type data and compares the sepsis medication list, and displays the sepsis reference drug list message through the display module. A sepsis care feedback system as described in claim 1, wherein the database stores a current time and a first time period; wherein the processing module further includes an emergency and critical care test unit; when the processing module receives the confirmed sepsis signal, the emergency and critical care test unit generates a critical care test table corresponding to the current time, and displays the critical care test table through the display module, and, each time the first time period passes, the emergency and critical care test unit regenerates the critical care test table; wherein the critical care test table has a plurality of mandatory test items and a plurality of optional test items for selection. A sepsis care feedback system as described in claim 1, wherein the database stores an infusion time threshold and a medication time threshold, and the processing module further includes an infusion reminder unit and a medication reminder unit; wherein, after the processing module generates the sepsis reference medication order message, and each time the infusion time threshold is passed, the infusion reminder unit generates an infusion reminder message, and displays the infusion reminder message through the display module; wherein, after the processing module generates the sepsis reference medication order message, and each time the medication time threshold is passed, the medication reminder unit generates a medication reminder message, and displays the medication reminder message through the display module. The sepsis care feedback system as described in claim 3, wherein the display module further displays a medication execution time field; wherein, when the medication reminder unit receives a medication execution time input into the medication execution time field from the human-machine interaction module, the medication reminder unit generates and displays the medication reminder message every time the medication execution time threshold is passed, starting from the medication execution time. A sepsis care feedback system as described in claim 1, wherein the database stores a current time and a second time period, and the processing module further includes a personalized feedback monitoring unit; wherein, after the processing module generates the sepsis reference medication order message, the personalized feedback monitoring unit starts to count an emergency room retention time, and generates a combined care table, and displays the combined care table and the emergency room retention time through the display module; wherein, the combined care table includes at least one care action to be performed at each interval of the second time period generated according to the sepsis reference medication order message; When the personalized feedback monitoring unit receives a completed care message from the human-machine interaction module, the personalized feedback monitoring unit generates a completed care message, and displays the completed care message through the display module. A sepsis care feedback system as described in claim 5, wherein the database stores a care time; wherein, at each interval of the second time period, the personalized feedback monitoring unit counts down the care time; wherein, before the countdown of the care time returns to zero, if the personalized feedback monitoring unit does not receive the executed care message from the human-machine interaction module, the personalized feedback monitoring unit generates a timeout message, and displays the timeout message through the display module. A sepsis care feedback system as described in claim 5, wherein the database stores a plurality of care note messages, and the care note messages correspond to the at least one care action; wherein, when the personalized feedback monitoring unit receives a signal for selecting the at least one care action from the human-machine interaction module, the personalized feedback monitoring unit displays one of the care note messages corresponding to the at least one care action through the display module. A sepsis care feedback system as described in claim 7, wherein the database stores a lactate value threshold and a plurality of care attention messages, and the processing module further includes an abnormal blood draw tracking reminder unit; Wherein, when the at least one care action is a first blood draw, and the abnormal blood draw tracking reminder unit determines that a lactate value in the blood is higher than the lactate value threshold, the abnormal blood draw tracking reminder unit notifies the personalized feedback monitoring unit to display one of the care attention messages corresponding to the at least one care action of a second blood draw through the display module. A sepsis care feedback system as described in claim 1, wherein the database stores a cardiac output threshold or a water-acid value threshold, and the processing module further includes a hemodynamic monitor evaluation unit, and the physiological information includes a cardiac output or a water-acid value; Wherein, when the hemodynamic monitor evaluation unit determines that the cardiac output of one of the patient data is less than the cardiac output threshold or the water-acid value of one of the patient data is greater than the water-acid value threshold, the hemodynamic monitor evaluation unit generates a hemodynamic abnormality message, and displays the hemodynamic abnormality message through the display module. The sepsis care feedback system as described in claim 1, wherein the processing module first searches, collects, filters and integrates the information of one of the patient data through the sepsis detection model, and then uses the sepsis detection model to determine whether the physiological information of one of the patient data meets the sepsis detection standard; wherein the sepsis detection model uses text recognition, phrase recognition, and code recognition to search and collect information related to one of the patient data from the database, then filters information not related to sepsis, and then integrates the remaining information related to sepsis.

Images