[go: up one dir, main page]

WO2023229279A1 - Procédé de détermination de l'âge à l'aide d'un microbiome - Google Patents

Procédé de détermination de l'âge à l'aide d'un microbiome Download PDF

Info

Publication number
WO2023229279A1
WO2023229279A1 PCT/KR2023/006593 KR2023006593W WO2023229279A1 WO 2023229279 A1 WO2023229279 A1 WO 2023229279A1 KR 2023006593 W KR2023006593 W KR 2023006593W WO 2023229279 A1 WO2023229279 A1 WO 2023229279A1
Authority
WO
WIPO (PCT)
Prior art keywords
age
information
prediction
microbiome
model
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/KR2023/006593
Other languages
English (en)
Korean (ko)
Inventor
황보경
윤성훈
이사도
황승진
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
LG H&H Co Ltd
Original Assignee
LG Household and Health Care Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Priority claimed from KR1020220064773A external-priority patent/KR102903988B1/ko
Application filed by LG Household and Health Care Ltd filed Critical LG Household and Health Care Ltd
Publication of WO2023229279A1 publication Critical patent/WO2023229279A1/fr
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • C—CHEMISTRY; METALLURGY
    • C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
    • C12Q—MEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEIC ACIDS OR MICROORGANISMS; COMPOSITIONS OR TEST PAPERS THEREFOR; PROCESSES OF PREPARING SUCH COMPOSITIONS; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL OR ENZYMOLOGICAL PROCESSES
    • C12Q1/00—Measuring or testing processes involving enzymes, nucleic acids or microorganisms; Compositions therefor; Processes of preparing such compositions
    • C12Q1/68—Measuring or testing processes involving enzymes, nucleic acids or microorganisms; Compositions therefor; Processes of preparing such compositions involving nucleic acids
    • C12Q1/6876—Nucleic acid products used in the analysis of nucleic acids, e.g. primers or probes
    • C12Q1/6888—Nucleic acid products used in the analysis of nucleic acids, e.g. primers or probes for detection or identification of organisms
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B40/00—ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B50/00—ICT programming tools or database systems specially adapted for bioinformatics
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients

Definitions

  • the present invention relates to a method for determining age using the microbiome.
  • the composition of the microbiome varies depending on age and health status, and changes in the information of the microbiome in the body cause aging and can be a signal of aging and various diseases.
  • microorganisms can be used to predict and manage the possibility of premature aging before obvious aging and disease appear.
  • Korean Patent Publication No. 10-2020-0054203 relates to a disease-related microbiome characterization process and discloses awareness of generating skin-related characterization models. Although the user's skin condition is diagnosed through genetic information and microbiome analysis, it does not disclose the use of this to estimate the user's age.
  • Korean Patent Publication No. 10-2021-0007512 relates to a cosmetics matching service system using a novel skin diagnosis method, which diagnoses the user's skin condition through genetic information and microbiome analysis. This does not disclose estimating the user's age.
  • U.S. Patent Publication No. 2021-0207201 discloses obtaining genes from skin cells of each actual age and learning them in order to estimate the age of the skin, but by learning microbiome information as learning data to create a learned model. , there is no awareness of judging age.
  • Patent Document 1 Korean Patent Publication No. 10-2020-0054203
  • Patent Document 2 Korean Patent Publication No. 10-2021-0007512
  • Patent Document 3 U.S. Patent Publication No. 2021-0207201
  • the present invention was created to solve the above problems.
  • the present invention builds a model for determining age based on the composition of the microbiome for each age group, but selects microorganisms that show a clear correlation with age judgment as variables to increase the accuracy of the model for determining age. am.
  • the present invention is intended to increase the accuracy of the model for determining age by selecting only the microbial variables that actually affect age prediction among numerous skin microorganisms through a multiple variable selection process.
  • the present invention is intended to output the age of a prediction target entity in a prediction terminal to which the generated age prediction model has been delivered.
  • the present invention is intended to output the age comparison result and customized prescription of the prediction target object at the prediction terminal to which the generated age prediction model has been delivered.
  • One embodiment of the present invention for solving the above problems includes a prediction terminal 100 that outputs the predicted age of a prediction target object using an age prediction model generated by a preset method; It includes, and the prediction terminal 100 includes an input unit 120 into which microbiome information of the prediction target entity is input; A processor 140 including an age calculation unit 141 that calculates the predicted age using the age prediction model; and an output unit 150 that outputs the calculated predicted age, wherein the age prediction model is a model learned using microbiome information as input data and age information as output data, and the age prediction model The model provides a system that outputs the predicted age of the prediction target entity when the microbiome information of the prediction target entity is input.
  • the age prediction model is generated in the server 200 and delivered to the prediction terminal 100, and the server 200 stores the subject's microbiome information and stores the age information.
  • the age prediction model is generated in the server 200 and delivered to the prediction terminal 100, and the server 200 includes the subject's microbiome information, subject information, and age information. stored memory 234; And the age prediction model is generated using the microbiome information, the subject information, and the age information, and the microbiome information and the age information are used to generate a plurality of models based on any one of the subject information. Classify into one of the classes, use the microbiome information classified for each model class as input data, and use the age information as output data to generate a plurality of learned age prediction models, which are then used as the plurality of models. and a learning processor 235 that generates a plurality of age prediction models by performing each class, and the subject information may include one or more of the subject's gender, lifestyle habits, and environmental information.
  • the prediction terminal 100 includes a memory 130 including a model storage unit 131 that receives and stores the age prediction model generated by the server 200; It further includes, the memory 130 further includes a prescription database 132 in which customized prescriptions according to the age comparison results are matched and stored in advance, and the processor 140 is configured to use the age calculation unit 141 an age comparison unit 142 that receives the calculated predicted age from and compares the predicted age with input age information of the prediction target object to generate the age comparison result value; And a prescription calculation unit 143 that receives the age comparison result value generated by the age comparison unit 142 and confirms a customized prescription corresponding to the transmitted age comparison result value in the prescription database 132. Additionally, the output unit 150 may further output the age comparison result and the customized prescription.
  • the prediction terminal 100 includes a memory 130 including a model storage unit 131 that receives and stores the age prediction model generated by the server 200; Further comprising, the memory 130 further includes a prescription database 132 in which customized prescriptions according to age comparison results, gender, lifestyle, and environmental information are matched and stored in advance, and the prediction terminal 100 ), the prediction target entity information is further input, and the processor 140 receives the calculated predicted age from the age calculation unit 141, and compares the input age information of the prediction target entity with the predicted age to compare the age.
  • An age comparison unit 142 that generates a result value; and a prescription calculation unit that receives the age comparison result generated by the age comparison unit 142 and confirms a customized prescription corresponding to the transmitted age comparison result and prediction target entity information in the prescription database 132. (143); wherein the output unit 150 further outputs the age comparison result and the customized prescription, and the prediction target entity information includes gender, lifestyle, and environmental information of the prediction target entity. Any one or more may be included.
  • the learning processor 235 uses the age information input to the prediction terminal 100 as output data, and the microbiome of the prediction target entity input to the prediction terminal 100. It may further include an additional learning unit 235d that additionally trains the age prediction model using information as input data.
  • the processor 233 generates a first data set pairing the microbiome information and the subject's age information, and divides the first data set into the first data set based on the age information. 1 Generate class information that classifies the data set into one age class among a plurality of age classes, and generate a second data set that pairs the first data set with the class information that classified the first data set.
  • the learning processor 235 includes a first variable selection unit 235a that generates a primary variable data set by selecting primary variables using a first preset method using the second data set; a second variable selection unit 235b that selects secondary variables using the primary variable data set using a second preset method to generate a secondary variable data set; and a model construction unit 235c that generates the age prediction model using the secondary variable data set, wherein the secondary variable data set includes selected microbiome information and age information, and the model construction unit 235c The unit 235c may generate the learned age prediction model using the selected microbiome information as input data and the age information as output data.
  • the first preset method may be a linear discriminant analysis (LEfSe) method
  • the second preset method may be a Boruta method based on RandomForest.
  • it further includes a data providing server 300 in which microbiome information is stored, and when subject identification information is input, the server 200 sends the subject identification information to the data providing server 300.
  • the data providing server 300 transmits microbiome information corresponding to the subject identification information to the server 200, and the prediction terminal 100 determines the prediction target when the prediction target entity identification information is input.
  • Entity identification information may be transmitted to the data providing server 300, and the data providing server 300 may transmit microbiome information corresponding to the prediction target entity identification information to the prediction terminal 100.
  • a method of outputting a predicted age using a prediction terminal includes the steps of: (a) inputting microbiome information of a prediction target entity to the prediction terminal 100; And (b) the prediction terminal 100 calculates the predicted age by inputting the microbiome information into an age prediction model pre-stored in the prediction terminal 100 or the server 200, and outputs the calculated predicted age. step; It includes, and provides a method in which the age prediction model is a model learned using microbiome information as input data and age information as output data.
  • a method of outputting predicted age and customized prescription using a prediction terminal includes the steps of: (c) inputting microbiome information and age information of the prediction target entity to the prediction terminal 100; (d) the prediction terminal 100 calculating a predicted age by inputting the microbiome information input in step (c) into an age prediction model previously stored in the prediction terminal 100 or the server 200; (e) the prediction terminal 100 comparing the calculated predicted age with the age information input in step (c) and calculating an age comparison result; and (f) the prediction terminal 100 confirming a customized prescription using the age comparison result and outputting the calculated age comparison result and the customized prescription; It includes, and provides a method in which the age prediction model is a model learned using microbiome information as input data and age information as output data.
  • step (c) inputting prediction target entity identification information into the prediction terminal 100 and transmitting the prediction target entity identification information to the data providing server 300; and (c2) when the prediction target entity identification information is transmitted, the data providing server 300 checks the microbiome information of the prediction target entity corresponding to the prediction target entity identification information, and sends the microbiome information to the prediction target entity.
  • transmitting to the prediction terminal 100 and inputting the microbiome information into the prediction terminal 100 may include.
  • the age prediction model is pre-generated in the server 200, and after step (e), the server 200 determines the microbiome of the prediction target entity input to the prediction terminal 100. It may further include the step of additionally training the age prediction model by using the information as input data and the age information of the prediction target object input to the prediction terminal 100 as output data.
  • the prediction terminal 100 contains microbiome information, age information, and prediction target entity information. input steps; (h) The prediction terminal 100 checks the prediction target entity information and, based on the prediction target entity information, determines one of a plurality of age prediction models pre-stored in the prediction terminal 100 or the server 200.
  • the prediction terminal 100 calculates a predicted age by inputting the microbiome information input in step (c); (i) the prediction terminal 100 compares the calculated predicted age with the age information input in step (c), and calculates an age comparison result; and (j) the prediction terminal 100 confirming a customized prescription using the age comparison result value and the prediction target entity information, and outputting the calculated age comparison result value and the customized prescription;
  • the plurality of age prediction models are models that are classified and learned according to any one of the subject information, each model is learned with microbiome information as input data and age information as output data.
  • the subject information and the prediction target entity information each include one or more of the age, gender, lifestyle, and environmental information of the subject and the prediction target entity.
  • step (g) inputting prediction target entity identification information to the prediction terminal 100 and transmitting the prediction target entity identification information to the data providing server 300; and (g2) when the prediction target entity identification information is transmitted, the data providing server 300 checks the microbiome information of the prediction target entity corresponding to the prediction target entity identification information, and sends the microbiome information to the prediction target entity.
  • transmitting to the prediction terminal 100 and inputting the microbiome information into the prediction terminal 100 may include.
  • the plurality of age prediction models are pre-generated in the server 200, and after step (j), the server 200 uses the microprocessor of the prediction target object input to the prediction terminal 100. It may further include the step of additionally training the age prediction model using biome information as input data and age information of the prediction target entity input to the prediction terminal 100 as output data.
  • a method of generating an age prediction model includes: (k) the processor 233 receiving a subject's microbiome information and the subject's age information; (l) generating, by the processor 233, a first data set consisting of a pair of the microbiome information and the subject's age information; (m) The processor 233 generates class information that classifies the first data set into one age class among a plurality of age classes, and second data pairs the first data set and the class information.
  • the learning processor 235 receives the second data set and selects strains in which the difference in microbial content between the plurality of age classes in the second data set is greater than or equal to a preset content difference or shows a statistically significant difference. Confirming and generating a primary variable data set including the content of the corresponding strain; (o) the learning processor 235 identifying a strain of high importance in the primary variable data set and generating a secondary variable data set including the content of the strain; and (p) the learning processor 235 constructing an age prediction model using the secondary variable data set.
  • the age prediction model constructed in step (p) includes the microbiome information. When you input, it provides a method, a model that outputs the predicted age.
  • the step (k) further includes (k1) the processor 233 receiving the subject's microbiome information and the subject's age information, and further receiving the subject's information
  • step (l) the processor 233 checks the input subject information, and selects the microbiome information and the age information based on any one of the subject information among a plurality of model classes. Classifying as one, generating a first data set consisting of a pair of the microbiome information and the age information classified into one of the plurality of model classes, respectively, for each of the plurality of model classes It further includes generating a plurality of first data sets by generating each first data set, wherein step (m) includes,
  • Step (m1) The processor 233 generates class information classifying each of the plurality of first data sets into one of the plurality of age classes, and the plurality of first data sets for each of the plurality of first data sets. generating a plurality of second data sets pairing one of the data sets with class information;
  • Step (n) includes: (n1) the learning processor 235 receiving the plurality of second data sets and generating the plurality of primary variable data sets;
  • the learning processor 235 identifies a strain of high importance in the primary variable data set and generates a secondary variable data set containing the content of the strain.
  • step (p) includes, (p1) the learning processor 235 generating the plurality of secondary variable data sets.
  • step (p1) includes, (p1) the learning processor 235 generating the plurality of secondary variable data sets.
  • the model outputs the predicted age, and the subject information and the predicted individual information may include one or more of the age, gender, lifestyle, and environmental information of the subject and the predicted individual information, respectively.
  • the learning processor 235 determines the difference in microbial content between the plurality of age classes using linear discriminant analysis (LEfSe) in the secondary data set. Confirming the strain showing and generating the primary variable data set; In the step (o) or (o1), the learning processor 235 performs a random forest (Random Forest) for the primary variable data set. It may include generating the secondary variable data set using Boruta analysis (RandomForest).
  • LfSe linear discriminant analysis
  • the learning processor 235 performs a random forest (Random Forest) for the primary variable data set. It may include generating the secondary variable data set using Boruta analysis (RandomForest).
  • the learning processor 235 may generate an age prediction model learned using a random forest.
  • step (k) includes: (k2) the processor 233 receiving subject identification information and transmitting the subject identification information to the data providing server 300; and (k3) when the subject identification information is transmitted, the data provision server 300 checks the subject's microbiome information corresponding to the subject identification information and transmits the microbiome information to the memory 234. Step; may include.
  • a computer program stored on a computer-readable recording medium is provided to execute the method according to any one of the above.
  • the present invention builds a model for determining age based on the composition of the microbiome for each age group, but can increase the accuracy of the model for determining age by selecting microorganisms that show a clear correlation with age judgment as variables.
  • the present invention can increase the accuracy of the model for determining age by selecting only the microbial variables that actually affect age prediction among numerous skin microorganisms through a multiple variable selection process.
  • the present invention can output the age of the prediction target entity from the prediction terminal to which the generated age prediction model has been delivered.
  • the present invention can output the age comparison result and customized prescription of the predicted object from the prediction terminal to which the generated age prediction model has been delivered.
  • Figure 1 is a diagram showing the network relationship between a server, a prediction terminal, and a data providing server.
  • Figure 2 is a diagram showing the control signal flow of a prediction terminal.
  • Figure 3 is a diagram showing the control signal flow of the server.
  • Figure 4 is a diagram for explaining how a data providing server transmits information to a prediction terminal and server, respectively.
  • FIG. 5 is a diagram illustrating generating a first age prediction model, transmitting it to a prediction terminal, and outputting the age as a first embodiment of the present invention.
  • Figure 6 is a second embodiment of the present invention, which illustrates generating a first age prediction model according to the first embodiment of the present invention, transmitting it to a prediction terminal, and outputting an age comparison result and a customized prescription. It is a drawing.
  • Figure 7 is a diagram for explaining the third embodiment of the present invention, generating a second age prediction model, transmitting it to a prediction terminal, and outputting the age comparison result and customized prescription.
  • Figure 8 is a diagram illustrating additional learning of an age prediction model using the age comparison result output according to the second embodiment of the present invention.
  • Figure 9 is a diagram illustrating additional learning of an age prediction model using the age comparison result output according to the third embodiment of the present invention.
  • Figure 10 is a diagram for explaining the information processing flow according to the first embodiment of the present invention.
  • Figure 11 is a diagram for explaining the information processing flow according to the second embodiment of the present invention.
  • Figure 12 is a diagram for explaining the information processing flow according to the third embodiment of the present invention.
  • FIG. 13 is a diagram illustrating a process for generating a first age prediction model by obtaining information necessary to create the first age prediction model from a subject.
  • FIG. 14 is a diagram illustrating a process for generating a second age prediction model by obtaining information necessary to create a second age prediction model from a subject.
  • Figure 15 is a flowchart explaining a method of outputting age in each prediction terminal.
  • Figures 16 and 17 are flowcharts for explaining a method of outputting age comparison results and customized prescriptions in a prediction terminal.
  • Figures 18 and 19 are flowcharts for explaining a method of generating an age prediction model according to the present invention.
  • Figure 20 is a diagram for explaining the process of variable selection through Boruta analysis in the method according to the present invention.
  • Figure 21 is a diagram illustrating the process of selecting a variable combination with the highest judgment accuracy and learning it to generate a prediction model constructed in the method according to the present invention.
  • FIG. 22 is a diagram illustrating the results of verifying the accuracy of the prediction model built in FIG. 21.
  • Figure 23 is a diagram for explaining a comparative example, and explains the suitability of a prediction model built by selecting variables through LEfSe analysis and learning them.
  • FIG. 24 is a diagram illustrating the results of verifying the accuracy of the prediction model built in FIG. 23.
  • prediction target entity refers to any entity, and the microbiome information and age information of the prediction target entity may be input to the prediction terminal 100, and the age of the prediction target entity may be output.
  • subject refers to an entity for learning a model
  • the subject's microbiome information and age information are input to the server 200 and used for model learning.
  • microbiome refers to microorganisms and their genes living in the human environment.
  • gene may be the full-length sequence of a microorganism, or may be the entire or partial gene of a specific region, for example, 16s rRNA and/or 18s rRNA, for phylogenetic analysis of microorganisms.
  • Microbiome information refers to the types of microorganisms present on the skin, especially the scalp, and/or the population ratio of the microorganisms in the total flora.
  • microorganisms refer to eukaryotes, which are made up of cells with a nucleus surrounded by a nuclear membrane, such as protozoa, algae, and fungi, and prokaryotes, which are made up of cells with no nuclear membrane and no distinct nucleus, such as eubacteria and archaic bacteria. It includes viruses and viroids that have substances but do not have a translation device or metabolic function, so they are difficult to see as cells.
  • strain refers to a purely isolated and cultured microorganism with a genetically identical nucleus.
  • importance means the number of standard deviations from the mean of a data point. For example, whether a true strain is more important than the best shadow feature means whether the strain has a higher number of standard deviations than the maximum number of standard deviations of the shadow feature.
  • age information refers to the actual age at the time the microbiome information was measured or pre-classified age information (e.g., teenagers, 20s, 30s, 40s, 50s, 60s, etc.) it means.
  • age information refers to the subject's actual age according to the resident registration number, and may be based on "full age,” but is not limited thereto.
  • predicted age refers to the age predicted through an age prediction model, and may not be the same as age information.
  • lifestyle habits refers to various lifestyle habits such as eating habits, washing habits, half-bathing habits, habits that stimulate the skin, waking up time, sleeping time, working time, and exercise habits including type and intensity of exercise and exercise time. All may be included and are not limited to the above.
  • environmental information may include temperature and humidity information in multiple spaces where the prediction target entity resides and is located, but is not limited to the above.
  • the device according to the present invention includes a prediction terminal 100 and a server 200.
  • the prediction terminal 100 receives a pre-learned age prediction model from the server 200 and predicts the age progression of the prediction target object using the age prediction model.
  • the server 200 may generate an age prediction model using various learning data, as will be described later, and this may configure each embodiment.
  • the server 200 selects variables through a preset variable selection method and creates an age prediction model using the selected variables, thereby increasing the accuracy of the age prediction model.
  • a query input to the prediction terminal 100 may be transmitted to the server 200 through a network, and may be input directly to the prediction terminal 100 or from the data providing server 300 through a microphone.
  • Biome information may be transmitted to the server 200, and the server 200 may calculate the predicted age using the generated age prediction model, and the calculated predicted age may be transmitted back to the prediction terminal 100 and output. there is.
  • a data providing server 300 may be further included, and the data providing server 300 may store the subject's microbiome information/microbiome information of the prediction target entity.
  • subject identification information is transmitted from the server 200 to the data provision server 300, and the data provision server 300 provides microbiome information corresponding to the transmitted subject identification information. can be transmitted to the server 200.
  • identification information of the prediction target object is transmitted from the prediction terminal 100 or the server 200 to the data providing server 300, and the data providing server 300 provides the microdata of the prediction target object corresponding to the transmitted identification information.
  • Ohm information can be transmitted to the prediction terminal 100 or the server 200.
  • the subject's microbiome information/microbiome information of the prediction target object stored in the data providing server 300 is transmitted to the server 200 and the prediction terminal 100 and used.
  • the subject's microbiome information/microbiome information of the prediction target entity is not limited to being stored in a separate data provision server 300.
  • the subject's microbiome information may be collected by the collection device 220 and directly input into the server 200.
  • the subject's microbiome information/microbiome information of the prediction object may be input to the server 200 and the prediction terminal 100 and stored, respectively.
  • the method by which microbiome information is input to each server 200 and the prediction terminal 100 is not limited to what is shown and described.
  • the prediction terminal 100 can predict the age of the prediction target entity by communicating with the server 200, and the data providing server 300 communicates with the prediction terminal 100 and the server 200, respectively, to obtain the subject's microbiome information. / Microbiome information of the prediction target object can be transmitted.
  • the prediction terminal 100 is a device that can predict the age of a prediction target entity, and receives the age prediction model generated by the server 200, which will be described later, to predict the age of the prediction target entity.
  • the prediction terminal 100 includes a communication unit 110, an input unit 120, a memory 130, a processor 140, an output unit 150, a control unit 160, a power supply unit 170, and an interface unit 180. do.
  • the communication unit 110 is configured for communication with an external device, and data may be transmitted and received with an external device through the communication unit 110.
  • the communication unit 110 communicates with external information and transmits information to the input unit 120.
  • the communication unit 110 may receive one or more information among prediction target entity information and age information of a specific entity.
  • the communication unit 110 may receive microbiome information of the prediction target entity from the data providing server 300.
  • the input unit 120 may receive commands or data to be used for components (e.g., processor, learning processor) of the prediction terminal 100 from outside of the prediction terminal 100 (e.g., a user).
  • the input unit 120 may include, for example, a microphone, mouse, keyboard, keys (eg, buttons), or a digital pen (eg, stylus pen).
  • the prediction target entity information, microbiome information, and age information input to the input unit 120 are transmitted to the processor 140, which will be described later, through the communication unit 110.
  • the memory 130 can store various data used in the prediction terminal 100.
  • Data may include, for example, input data or output data for software and instructions related thereto.
  • the memory 130 may store both information input to the prediction terminal 100 and information generated by the prediction terminal 100.
  • the memory 130 includes a model storage unit 131 and a prescription database 132.
  • the model storage unit 131 stores the age prediction model generated by the server 200.
  • prescription database 132 appropriate customized prescriptions are stored in advance according to age and age comparison results.
  • customized prescriptions according to age, gender, lifestyle, and environmental information may be stored in the prescription database 132.
  • Customized prescriptions can refer to prescription products and prescription methods.
  • prescription products may include, but are not limited to, drugs or cosmetics that are pre-matched and stored according to predicted age and recommended to the user.
  • the prescription method may include, but is not limited to, matching and saving in advance according to the predicted age, and lifestyle and exercise patterns recommended to the user.
  • a customized prescription may mean exercising more than a few times a week if the age comparison result described later is positive (+). If the age comparison result is negative (-), customized prescription may mean maintaining current habits, but is not limited to this.
  • prescription products or prescription methods according to age stored in the prescription database 132 may be continuously updated.
  • Microbiome information may be obtained from a sample collected from the skin of a subject or prediction target entity, and may be information collected through the collection device 220.
  • the way microbiome information is collected is not limited to a specific method.
  • microbiome information is personal information and is stored in the data provision server 300, and when prediction target entity identification information is transmitted, microbiome information corresponding thereto can be transmitted to the prediction terminal 100.
  • microbiome information can be obtained from samples collected from specific areas of the skin, but is not limited to this. Microbiome information can also be collected using microorganisms in a specific area of the skin that is to be determined.
  • the prediction target entity information may include one or more of the prediction target entity's gender, lifestyle, and environmental information.
  • Lifestyle habits and environmental information may be collected from survey results of the prediction target entity, but the method in which lifestyle habits and environmental information is collected is not limited to this.
  • the prediction target entity information is not limited to the above, and of course may further include information that affects age prediction.
  • the age prediction model is a model in which a data set consisting of pairs of microbiome information and age information is learned, and microorganisms that are unrelated to age prediction are excluded from the microbiome information and a selected variable data set is used. .
  • variable data set was selected by excluding microorganisms unrelated to age prediction, and may be information selected through Boruta analysis based on linear discriminant analysis (LEfSe) and Random Forest. .
  • the processor 140 may perform a data processing or calculation function and, as at least part of the data processing or calculation function, store instructions or data received from other components in a volatile memory, and store the instructions or data stored in the volatile memory. can be processed, and the resulting data can be stored in non-volatile memory.
  • the processor 140 may include an age calculation unit 141, an age comparison unit 142, and a prescription calculation unit 143.
  • the processor 140 includes an age calculation unit 141, and according to the second and third embodiments, the processor 140 includes an age calculation unit 141, an age comparison unit 142, and Includes both prescription calculation unit 143.
  • the age calculation unit 141 uses the age prediction model stored in the memory 130, inputs the microbiome information of the prediction target object to the age prediction model, and calculates the predicted age.
  • the age comparison unit 142 compares the predicted age calculated by the age calculation unit 141 with the age information of the prediction target object input to the prediction terminal 100.
  • the age comparison unit 142 may calculate the age comparison result value by comparing the predicted age (predicted value) calculated from the age prediction model with the age information (actual value) of the prediction target object.
  • the age comparison result value may mean a positive or negative number, which is the difference between the predicted age and age information, but is not limited thereto. For example, if the age information is 32 years old, but the predicted age is 29 years old, the age comparison result value may be -3.
  • the prescription calculation unit 143 calculates a customized prescription corresponding to the age comparison result value according to the age comparison result value calculated by the age comparison unit 142.
  • the prescription calculation unit 143 may calculate a customized prescription according to the age comparison result using a prescription product or prescription method according to age previously stored in the prescription database 132. For example, the prescription calculation unit 143 can calculate how many grams or more times a certain drug will be used based on a specific age comparison result.
  • Customized prescriptions can vary depending on whether the age comparison result value is positive (if the predicted age is older than the actual age) or negative (if the predicted age is younger than the actual age). Even if the result is the same positive number, the prescription will vary depending on the value. You can.
  • the output unit 150 may output information processed or calculated in the prediction terminal 100 to the outside, and a display and speaker may be included here.
  • the output unit 150 outputs the results calculated by the processor 140.
  • the output unit 150 may output one or more of predicted age, age comparison results, and customized prescriptions.
  • the control unit 160 may control one or more other components (eg, hardware or software components) of the prediction terminal 100 connected to the processor by executing software, for example.
  • the control unit 160 controls the communication unit 110, input unit 120, memory 130, processor 140, output unit 150, power supply unit 170, and interface unit 180 of the prediction terminal 100. can do.
  • the power supply unit 170 supplies power to the prediction terminal 100.
  • the power supply unit 170 may receive power from an external power source and supply power to the prediction terminal 100.
  • the interface unit 180 may display the age output from the output unit 150 through an interface.
  • the interface unit 180 is not limited to a specific type or form.
  • the server 200 includes a prediction model generating device 230.
  • the server 200 may also be implemented in the form of a computing device like the prediction terminal 100 described above, and may include components included in the prediction terminal 100 as is or at least some of them. Descriptions of the general functions of each component overlapping with the prediction terminal 100 will be omitted.
  • the prediction model generating device 230 includes a communication unit 231, an input unit 232, a processor 233, a memory 234, and a learning processor 235.
  • the communication unit 231 communicates with the outside and transmits the relevant information to the input unit 232.
  • the input unit 232 may transmit subject information, microbiome information, and age information input through the communication unit 231 to the integrated database 234a, which will be described later.
  • the communication unit 231 receives microbiome information from the data providing server 300, but the method of transmitting microbiome information is not limited to a specific method.
  • the processor 233 receives the microbiome information and the corresponding age information stored in the memory 234 and generates the microbiome information and age information as a pair of first data sets.
  • the processor 233 generates class information that classifies the first data set into one age class among a plurality of age classes based on age, and generates class information that includes the class information and the corresponding first data set. Create a data set.
  • the plurality of age classes may be classes classified based on age differences of 10 years, such as teenagers, 20s, 30s, and 40s, but are not limited thereto.
  • the processor 233 may transmit the generated second data set to the learning processor 235.
  • the processor 233 may use the subject information stored in the memory 234 to classify microbiome information-age information using the subject information and generate a plurality of first data sets and second data sets. there is. A detailed description of this will be provided later in the description of the third embodiment.
  • the memory 234 may store both information input to the server 200 and information generated by the server 200. Additionally, the prediction model generated through learning may also be stored in the memory 234 of the server 200.
  • Memory 234 includes an integrated database 234a.
  • the integrated database 234a stores one or more of age information and subject information transmitted from the input unit 232.
  • Subject information may include one or more of the subject's gender, lifestyle habits, and environmental information.
  • Lifestyle habits and environmental information refer to the results of a survey conducted by the subject in advance.
  • the subject information is not limited to the above, and may further include other information about the subject that affects age prediction.
  • the integrated database 234a contains the microbiogram transmitted from the input unit 232. Ohmic information may be stored, and a first data set and a second data set generated by the processor 233 may be stored.
  • the learning processor 235 can select variables to create a variable data set using the second data set generated by the processor 233, and use this to build an age prediction model.
  • the learning processor 235 may include a hardware structure specialized for processing artificial intelligence models.
  • the learning processor 235 includes a first variable selection unit 235a, a second variable selection unit 235b, and a model construction unit 235c.
  • the first variable selection unit 235a selects primary variables using the second data set generated by the processor 233 using a first preset method to generate a primary variable data set.
  • the first preset method may mean linear discriminant analysis, but is not limited to a specific method.
  • the second variable selection unit 235b receives the primary variable data set from the first variable selection unit 235a, uses the primary variable data set to select secondary variables using a second preset method, and selects the secondary variable as a selection microbiode. Generate ohmic information and create a secondary variable data set.
  • the second preset method may mean Boruta analysis, but is not limited to a specific method.
  • the model building unit 235c receives the secondary variable data set from the second variable selection unit 235b and creates an age prediction model using the secondary variable data set.
  • the learning processor 235 further includes an additional learning unit 235d, and can additionally learn an age prediction model using age information input to the prediction terminal 100.
  • the additional learning unit 235d may additionally learn an age prediction model by using the microbiome information of the prediction object as input data and the age information input from the prediction terminal 100 at that time as output data.
  • the additionally learned age prediction model can increase prediction accuracy as it is further learned using the input age information.
  • the first age prediction model is a model learned using the subject's microbiome information and age information as learning data, and the processor 140 of the prediction terminal 100 uses the age calculation unit 141. Includes.
  • the predicted age of the prediction target entity calculated according to the first age prediction model may be output.
  • the communication unit 231 receives it and transmits it to the input unit 232.
  • the server 200 receives subject identification information.
  • the server 200 may receive microbiome information from the collection device 220, but may also receive microbiome information from the data providing server 300.
  • the subject identification information is not limited to specific information as long as it is information that can identify the subject in question in the data provision server 300.
  • the server 200 transmits the generated subject identification information to the data providing server 300.
  • the data providing server 300 receives the subject identification information and transmits the subject's microbiome information corresponding thereto to the server 200, and the communication unit 231 receives this and transmits it to the input unit 232.
  • the subject's microbiome information and age information transmitted to the input unit 232 are stored in the integrated database 234a.
  • the above process is repeated, and the microbiome information of multiple subjects and age information including actual age are stored in the integrated database 234a.
  • the processor 233 When the microbiome information and actual age are stored in the integrated database 234a, the processor 233 generates a first data set consisting of a pair of the subject's microbiome information and the subject's age information. In addition, the processor 233 generates class information that classifies the first data set into one of a plurality of age classes based on age, and second data that pairs the first data set and the class information. Create a set.
  • the learning processor 235 learns the plurality of second data sets stored in the memory 234 to generate a first age prediction model.
  • the first age prediction model uses microbiome information as input and predicted age as output.
  • the communication unit 231 transmits the generated first age prediction model to the prediction terminal 100.
  • the communication unit 110 of the prediction terminal 100 receives this and transmits it to the input unit 120, and the input unit 120 transmits the first age prediction model to the model storage unit 131 to store the first age prediction model in the model storage unit 131. 1
  • the age prediction model is saved.
  • the prediction terminal 100 receives prediction target entity identification information.
  • the prediction target entity identification information is not limited to specific information as long as it is information that allows the data providing server 300 to identify the prediction target entity.
  • the data providing server 300 receives the prediction target entity identification information and transmits the microbiome information of the prediction target entity including the corresponding information to the prediction terminal 100, and the communication unit 110 receives this, It is transmitted to the input unit 120.
  • the input unit 120 inputs the microbiome information of the prediction target entity to the age calculation unit 141 of the processor 140.
  • the age calculation unit 141 calculates the predicted age of the prediction target object input to the prediction terminal 100 using the stored first age prediction model.
  • the output unit 150 receives the age from the age calculation unit 141 and outputs the predicted age of the prediction target entity.
  • the first age prediction model is generated in the same manner as described in the above-described first embodiment, and the processor 140 of the prediction terminal 100 performs the age calculation unit 141 and the age comparison. It further includes a unit 142 and a prescription calculation unit 143.
  • the prediction terminal 100 when the microbiome information of the object to be predicted is input to the prediction terminal 100, the predicted age is output according to the first age prediction model, and the prediction terminal 100 uses this to output the predicted age. , the age comparison results of the predicted object and customized prescriptions can be output.
  • the process of generating the first age prediction model and storing the first age prediction model in the model storage unit 131 is the same as the first embodiment described above, and the detailed description thereof will replace the above.
  • the communication unit 110 receives it and transmits it to the input unit 120, and the input unit 120 calculates the age of the processor 140.
  • Microbiome information of the prediction target entity is input into the calculation unit 141.
  • the age calculation unit 141 calculates the age of the prediction target object using the stored first age prediction model.
  • the age comparison unit 142 compares the predicted age calculated by the age calculation unit 141 with the age information of the prediction target object input to the prediction terminal 100.
  • the age comparison unit 142 may change the predicted age output from the first age prediction model using a value calculated using the input age information of the prediction target object to generate an age comparison result value.
  • the age comparison result value may mean a value obtained by subtracting the predicted age from the age information of the prediction target entity, but is not limited thereto.
  • the prescription calculation unit 143 calculates a customized prescription corresponding to the age comparison result value according to the age comparison result value generated by the age comparison unit 142.
  • the prescription calculation unit 143 uses prescription products or prescription methods according to age pre-stored in the prescription database 132 to calculate or confirm a customized prescription corresponding to the age comparison result.
  • the output unit 150 receives the age comparison result and customized prescription from the processor 140 and outputs them.
  • the first age prediction model is additionally learned.
  • the prediction terminal 100 transmits the input age information (actual age information) and the input microbiome information of the prediction target entity to the server 200.
  • Age information and microbiome information transmitted to the server 200 are stored in the memory 234.
  • the microbiome information of the prediction target object transmitted to the server 200 and the input age information are transmitted to the processor 233, forming a data set pairing the microbiome information and age information of the prediction target object. It can be.
  • the additional learning unit 235d may additionally train the first age prediction model learned using the data set.
  • the second age prediction model is a model learned using the subject's microbiome information, age information, and subject information as learning data
  • the processor 140 of the prediction terminal 100 uses the age calculation unit (141), it includes an age comparison unit (142) and a prescription calculation unit (143).
  • the microbiome information and age information are classified using the subject information, and then the microbiome information and age classified using the subject information are generated. Learned through information.
  • a plurality of second age prediction models are generated by learning microbiome information and age information classified as subject information, respectively.
  • the corresponding predicted age when the microbiome information of the prediction target object is input to the prediction terminal 100, the corresponding predicted age can be output according to the second age prediction model, and the age comparison result value and customized prescription are provided. can be output.
  • the communication unit 231 receives the information and transmits it to the input unit 232.
  • the server 200 receives subject identification information of the subject.
  • the server 200 may receive microbiome information from the collection device 200, and the method by which microbiome information is input to the server 200 is not limited to a specific method.
  • the server 200 transmits the generated subject identification information to the data providing server 300.
  • the data providing server 300 receives the subject identification information and transmits the subject's microbiome information corresponding thereto to the server 200, and the communication unit 231 receives this and transmits it to the input unit 232.
  • the subject's microbiome information, age information, and subject information transmitted to the input unit 232 are stored in the integrated database 234a.
  • the above process is repeated, and the microbiome information, age information, and subject information of a plurality of subjects are stored in the integrated database 234a.
  • the learning processor 235 generates a second age prediction model that learns microbiome information and age information in pairs based on preset information among the input subject information, and repeats this for each type of preset information. Generate a plurality of second age prediction models.
  • a plurality of second age prediction models use microbiome information as input data, and learn age information and subject information as output data, respectively. As described above, they are classified into subject information and generated as respective prediction models.
  • preset information among subject information may mean gender.
  • a second age prediction model - male and a second age prediction model - female classified into male and female can be generated, respectively.
  • the communication unit 231 transmits the plurality of generated second age prediction models to the prediction terminal 100.
  • the communication unit 110 of the prediction terminal 100 receives this and transmits it to the input unit 120, and the input unit 120 transmits a plurality of second age prediction models to the model storage unit 131 to store the model storage unit 131.
  • a plurality of second age prediction models are stored.
  • the prediction terminal 100 receives prediction target entity identification information.
  • the data providing server 300 receives the prediction target entity identification information and transmits the microbiome information of the prediction target entity including the corresponding information to the prediction terminal 100, and the communication unit 110 receives this, It is transmitted to the input unit 120.
  • prediction target entity information is input to the prediction terminal 100.
  • the prediction target entity information may include one or more of the prediction target entity's gender, lifestyle habits, and environmental information.
  • the input unit 120 inputs the microbiome information of the prediction target object to the age calculation unit 141 of the processor 140, Microbiome information is input into one of the plurality of second age prediction models based on .
  • the age calculation unit 141 calculates the predicted age of the prediction target object using the stored second age prediction model.
  • the age comparison unit 142 compares the predicted age calculated by the age calculation unit 141 with the age information of the prediction target object input to the prediction terminal 100.
  • the age comparison unit 142 may calculate an age comparison result by comparing the predicted age output from the second age prediction model with the input age information.
  • the age comparison result value means the difference between the age information of the prediction target object and the predicted age. It means the difference between the predicted age and the age information of the prediction target object.
  • the prescription calculation unit 143 calculates a customized prescription corresponding to the age comparison result value according to the age comparison result value generated by the age comparison unit 142.
  • the prescription calculation unit 143 checks prescription products or prescription methods according to age previously stored in the prescription database 132 and calculates a customized prescription according to the age comparison result.
  • the customized prescription may mean a prescription product or prescription method determined by further using the prediction target entity information input to the prediction terminal 100.
  • customized prescription may include changing the behavior pattern or habits of the predicted object using the age comparison result and predicted object information.
  • the output unit 150 receives the predicted age, age comparison result, and customized prescription from the processor 140 and outputs them.
  • the prediction terminal 100 transmits the input microbiome information and age information of the prediction target entity to the server 200.
  • the microbiome information and age information of the prediction target object transmitted to the server 200 are transmitted to the processor 233, so that a data set pairing the microbiome information and age information of the prediction target object can be formed. there is.
  • the additional learning unit 235d may additionally train the second age prediction model learned using the data set, and the second age prediction model may be updated.
  • a method for generating an age prediction model according to the present invention includes a server 200.
  • the server 200 includes a prediction model generating device 230.
  • the prediction model generating device 230 includes a communication unit 231, an input unit 232, a processor 233, a memory 234, and a learning processor 235.
  • the processor 233 receives one or more of the subject's microbiome information and age information.
  • the processor 233 When age information is input, the processor 233 generates a first data set consisting of a pair of microbiome information and age information.
  • the processor 233 generates class information that classifies the first data set into one of a plurality of age classes based on age, and creates a second data set that pairs the first data set and the class information. creates .
  • the processor 233 labels each first data set as an age class among a plurality of classes according to the numerical value of age.
  • the first variable selection unit 235a identifies strains in the second data set in which the difference in microbial content between a plurality of age class information is greater than or equal to a preset content difference, and selects a primary variable including the content of the strain showing more than the preset content difference. Create a variable data set.
  • the first variable selection unit 235a can use Linear discriminant analysis Effect Size (LEfSe) analysis to identify strains in which the difference in microbial content is greater than or equal to a preset content difference or shows a statistically significant difference. Meanwhile, the first variable selection unit 235a has been described as selecting variables using Linear discriminant analysis Effect Size (LEfSe) analysis, but is not limited thereto.
  • LfSe Linear discriminant analysis Effect Size
  • statistically significant may mean a value at a level that is generally significant in statistics to those skilled in the art. For example, it can usually be defined by p-value.
  • a statistical difference means that it is repeated with a high probability when the same test is performed. For example, this may mean that it repeats with a 95%, 99%, or 99.9% probability.
  • the p-value of a statistically significant difference may be 5%, 1%, and 0.1%, respectively, but is not limited thereto.
  • the first variable selection unit 235a receives the second data set and tests whether the microbial content within different age classes in the second data set is statistically significantly differentially distributed (P-value ⁇ 0.05 ), and select strains in which the difference in microbial content between multiple age classes is distributed beyond the preset content difference.
  • an example method of the test that can be used includes the non-parametric factorial Kruskal-Wallis (KW) sum-rank test, but is not limited thereto.
  • the first variable selection unit 235a tests whether strains showing content differences between age classes have significant differences in pairwise comparisons between arbitrary subclasses (P-value ⁇ 0.05). That is, the first variable selection unit 235a checks whether there is more than a preset difference in content even when comparing subclasses whose microbial content among multiple age classes is less than the standard value.
  • the pairwise Wilcoxon test can be used as an example of the test that can be used, but is not limited thereto.
  • the first variable selection unit 235a builds an LDA model that ranks strains for strains with preset content differences between multiple age classes and calculates the effect size of each strain.
  • the effect size calculation method can be the average of a and b, where a is the difference between the averages of each class, and b is the class average when equal weight is given to the variability and discriminant power of the strain along the first linear discriminant axis. It means the difference value between
  • the LDA score refers to the linear discriminant analysis Effect Size result and the score for the quantitative predictor variable in log scale.
  • the first variable selection unit 235a may transmit the primary variable data set including the content of the identified strain to the second variable selection unit 235b.
  • the second variable selection unit 235b receives the primary variable data set from the first variable selection unit 235a.
  • the second variable selection unit 235b selects secondary variables using the primary variable data set using a second preset method to generate a secondary variable data set.
  • the second variable selection unit 235b re-identifies microorganisms that are unrelated to age classification among the types of microorganisms identified in the first variable selection unit 235a using a second preset method, and selects microbiome information. and generate a secondary variable data set containing selected microbiome information.
  • the second variable selection unit 235b transmits the secondary variable data set to the model construction unit 235c.
  • the second variable selection unit 235b may use a RandomForest-based variable importance analysis method, and more specifically, may use the Boruta algorithm to generate a secondary variable data set. , but is not limited to this.
  • the second variable selection unit 235b is described as selecting variables using Boruta analysis, but the method for selecting variables is not limited to this.
  • Boruta analysis determines the ranking of microorganisms that are important in determining age information among microorganisms with large statistically significant differences identified in LEfSe analysis.
  • the second variable selection unit 235b replicates the data set of strains selected in the first variable selection unit 235a and expands the data by mixing the values of each column to create a shadow variable (shadow feature).
  • the shadow variable refers to virtual data randomly generated using existing data.
  • the second variable selection unit 235b trains a random forest classifier using the expanded data set and evaluates the importance of each strain.
  • the evaluation method may be average reduction accuracy, average impurity reduction method, etc.
  • the second variable selection unit 235b checks whether the actual strain has a higher importance than the best shadow variable (i.e., whether the strain has a Z score higher than the maximum Z score of the shadow variable), and strains with high importance are selected. and stored, and strains deemed unimportant are continuously removed.
  • Z-score refers to the number of standard deviations from the mean of data points.
  • subject information is further used to generate a plurality of first data sets and second data sets for each piece of information included in the subject information, and each second age prediction model is generated. Since the learning method is the same as the first age prediction model except for generating multiple age prediction models using further subject information, detailed description of overlapping content will be omitted.
  • the processor 233 receives the subject's microbiome information and the subject's age information, and further receives subject information.
  • the processor 233 checks the input subject information and classifies the microbiome information and age information into one of a plurality of model classes based on one of the subject information.
  • a first data set consisting of a pair of microbiome information and age information is each generated to generate a plurality of first data sets.
  • the first data set may mean microbiome information-age information-model class.
  • the model class is a class that classifies one piece of information among the subject information. For example, it may mean a class that classifies the subject information using gender. In this case, this may mean classifying microbiome information and age information into either a female model class or a male model class.
  • the processor 233 generates class information that classifies each of the plurality of first data sets into one of the plurality of age classes, and each of the plurality of first data sets is classified into one of the plurality of first data sets and a class information. Create a plurality of secondary data sets pairing the information.
  • the learning processor 235 receives the plurality of second data sets and generates the plurality of primary variable data sets.
  • the learning processor 235 identifies strains of high importance in the primary variable data set, generates a secondary variable data set containing the content of the corresponding strain, and repeats this for each other primary variable data set to create a plurality of secondary variable data sets. Create a variable data set.
  • the learning processor 235 builds a plurality of second age prediction models using each of the plurality of secondary variable data sets.
  • the plurality of constructed second age prediction models are models that are classified and learned according to one of the subject information, and are models that output the predicted age when microbiome information of the prediction target object is input.
  • the subject information and the prediction target entity information include one or more of the age, gender, lifestyle, and environmental information of the subject and the prediction target entity, respectively.
  • Figure 20 shows variables selected using the Boruta algorithm in the method according to the invention.
  • Figure 20 illustrates further removing variables from the microbial types identified in the primary variable data set.
  • the second variable selection unit 235b removes variables that are not important in dividing the dependent variable, age, selects microorganisms of high importance that can be used in the model as variables, and generates a secondary variable data set.
  • the model building unit 235c builds an age prediction model using the secondary variable data set.
  • the secondary variable data set includes selected microbiome information, which is microbial information selected as variables in the first variable selection unit 235a and the second variable selection unit 235b, as input data, and , the predicted age at that time is included as output data.
  • the model building unit 235c may learn the age prediction model using a random forest, but the learning method of the model building unit 235c is not limited to this.
  • the model construction unit 235c randomly classifies the content data of the strain selected in the second variable selection unit 235b into a training set or a validation set. For example, it may be set to 80% training set and 20% validation set.
  • the model building unit 235c can learn a model using a training set and select the optimal variable combination with the highest Accuracy value for each model.
  • the model building unit 235c can build a classification model using the selected optimal variable combination and training set. Additionally, the model building unit 235c may evaluate and output the diagnostic correct classification rate using the validation set.
  • FIG. 22 is a diagram illustrating the results of an experiment verifying the accuracy of the prediction model built in FIG. 21.
  • Age diagnosis was performed on 30 subjects, and the diagnosis results of a total of 18 subjects matched the actual symptoms, so the age was diagnosed with a correct classification rate of 60%.
  • Figure 23 is a diagram for explaining a comparative example, and is a diagram for selecting variables only through LEfSe analysis, explaining the suitability of a prediction model built by learning it, and explaining the results of verifying the accuracy of the prediction model.
  • the age prediction model according to the present invention can predict age with a correct classification rate of 60% as a result of the verification experiment as described above, and is 16.7% better in age diagnosis than the conventional model learned using only LEfSe analysis. You can find out what is accurate.
  • the method according to an embodiment of the present invention described above may be implemented in the form of computer program instructions that can be executed through various computer means and recorded on a computer-readable medium.
  • the computer-readable medium may include program instructions, data files, data structures, etc., singly or in combination.
  • Program instructions recorded on the medium may be specially designed and constructed for the present invention or may be known and usable by those skilled in the art of computer software.
  • Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic media such as floptical disks.
  • program instructions include machine language code, such as that produced by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc.
  • the hardware devices described above may be configured to operate as one or more software modules to carry out the operations of the present invention, and vice versa.

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Physics & Mathematics (AREA)
  • Medical Informatics (AREA)
  • Chemical & Material Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • General Health & Medical Sciences (AREA)
  • Theoretical Computer Science (AREA)
  • Analytical Chemistry (AREA)
  • Databases & Information Systems (AREA)
  • Biophysics (AREA)
  • Proteomics, Peptides & Aminoacids (AREA)
  • Organic Chemistry (AREA)
  • Public Health (AREA)
  • Software Systems (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Biotechnology (AREA)
  • Epidemiology (AREA)
  • Evolutionary Computation (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Zoology (AREA)
  • Wood Science & Technology (AREA)
  • Evolutionary Biology (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioethics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Primary Health Care (AREA)
  • Mathematical Physics (AREA)
  • General Physics & Mathematics (AREA)
  • Immunology (AREA)
  • Microbiology (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Pathology (AREA)
  • Biomedical Technology (AREA)
  • Biochemistry (AREA)
  • Genetics & Genomics (AREA)

Abstract

La présente invention concerne un procédé et un système de prédiction de l'âge à l'aide d'un microbiome, un terminal de prédiction pouvant utiliser un modèle de prédiction de l'âge préalablement entraîné et pouvant délivrer un ou plusieurs des éléments suivants : l'âge, une valeur de résultat de comparaison d'âges et une prescription personnalisée. En outre, le modèle de prédiction de l'âge sélectionne, en tant que variables, des micro-organismes qui présentent une corrélation claire avec l'âge et ne sélectionne que des variables microbiennes qui ont un effet réel sur la détermination de l'âge parmi de nombreux micro-organismes par l'intermédiaire d'une pluralité de processus de sélection variables, ce qui permet de prédire l'âge avec davantage de précision.
PCT/KR2023/006593 2022-05-26 2023-05-16 Procédé de détermination de l'âge à l'aide d'un microbiome Ceased WO2023229279A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
KR10-2022-0064773 2022-05-26
KR1020220064773A KR102903988B1 (ko) 2022-05-26 마이크로바이옴을 이용한 나이 판단 방법

Publications (1)

Publication Number Publication Date
WO2023229279A1 true WO2023229279A1 (fr) 2023-11-30

Family

ID=88919601

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/KR2023/006593 Ceased WO2023229279A1 (fr) 2022-05-26 2023-05-16 Procédé de détermination de l'âge à l'aide d'un microbiome

Country Status (1)

Country Link
WO (1) WO2023229279A1 (fr)

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106202989A (zh) * 2015-04-30 2016-12-07 中国科学院青岛生物能源与过程研究所 一种基于口腔微生物群落获得儿童个体生物年龄的方法
US20200075127A1 (en) * 2017-07-25 2020-03-05 Deep Longevity Limited Aging markers of human microbiome and microbiomic aging clock

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106202989A (zh) * 2015-04-30 2016-12-07 中国科学院青岛生物能源与过程研究所 一种基于口腔微生物群落获得儿童个体生物年龄的方法
US20200075127A1 (en) * 2017-07-25 2020-03-05 Deep Longevity Limited Aging markers of human microbiome and microbiomic aging clock

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
FEDOR GALKIN, POLINA MAMOSHINA, ALEX ALIPER, EVGENY PUTIN, VLADIMIR MOSKALEV, VADIM N. GLADYSHEV, ALEX ZHAVORONKOV: "Human Gut Microbiome Aging Clock Based on Taxonomic Profiling and Deep Learning", ISCIENCE, CELL PRESS, US, vol. 23, no. 7, 26 June 2020 (2020-06-26), US , pages 101199, XP093112237, ISSN: 2589-0042, DOI: 10.1016/j.isci *
HUANG SHI, HAIMINEN NIINA, CARRIERI ANNA-PAOLA, HU REBECCA, JIANG LINGJING, PARIDA LAXMI, RUSSELL BAYLEE, ALLABAND CELESTE, ZARRIN: "Human Skin, Oral, and Gut Microbiomes Predict Chronological Age", MSYSTEMS, HIGHWIRE PRESS (FREE ACCESS), vol. 5, no. 1, 11 February 2020 (2020-02-11), pages e00630 - 19, XP093112234, ISSN: 2379-5077, DOI: 10.1128/mSystems.00630-19 *
SHEN JIE; ZHANG DAKE; LIANG BOYING: "Prediction of host age and sex classification through gut microbes based on machine learning", BIOCHEMICAL ENGINEERING JOURNAL, ELSEVIER, AMSTERDAM, NL, vol. 178, 3 December 2021 (2021-12-03), NL , pages 108280, XP086903385, ISSN: 1369-703X, DOI: 10.1016/j.bej.2021.108280 *

Also Published As

Publication number Publication date
KR20230164973A (ko) 2023-12-05

Similar Documents

Publication Publication Date Title
WO2020138624A1 (fr) Appareil de suppression de bruit et son procédé
WO2020111880A1 (fr) Procédé et appareil d'authentification d'utilisateur
WO2023080379A1 (fr) Appareil de génération d'informations d'apparition de maladie basé sur une corrélation temporelle à l'aide d'un score de risque polygénique et son procédé
WO2023191206A1 (fr) Système et procédé d'automatisation d'analyse de données d'exploration sur la base d'attributs variables
WO2023080766A1 (fr) Appareil pour générer des informations de mutation de gène à risque spécifique à une maladie à l'aide d'un modèle prs reposant sur une covariable variant dans le temps, et procédé associé
WO2018143707A1 (fr) Système d'evaluation de maquillage et son procédé de fonctionnement
WO2020060161A1 (fr) Système d'analyse statistique et méthode d'analyse statistique utilisant une interface conversationnelle
WO2021132851A1 (fr) Dispositif électronique, système de soins du cuir chevelu et son procédé de commande
WO2022010255A1 (fr) Procédé, système et support lisible par ordinateur permettant la déduction de questions approfondies destinées à une évaluation automatisée de vidéo d'entretien à l'aide d'un modèle d'apprentissage automatique
Talib et al. Fuzzy decision-making framework for sensitively prioritizing autism patients with moderate emergency level
WO2014021567A1 (fr) Procédé pour la fourniture d'un service de messagerie, et dispositif et système correspondants
WO2020218635A1 (fr) Appareil de synthèse vocale utilisant une intelligence artificielle, procédé d'activation d'appareil de synthèse vocale et support d'enregistrement lisible par ordinateur
WO2020209693A1 (fr) Dispositif électronique pour mise à jour d'un modèle d'intelligence artificielle, serveur, et procédé de fonctionnement associé
WO2020017827A1 (fr) Dispositif électronique et procédé de commande pour dispositif électronique
WO2020145571A2 (fr) Procédé et système de gestion d'un modèle d'évaluation automatique pour une vidéo d'entrevue et support lisible par ordinateur
WO2021162481A1 (fr) Dispositif électronique et son procédé de commande
WO2021086127A1 (fr) Dispositif concentrateur, système multi-dispositif comprenant le dispositif concentrateur et une pluralité de dispositifs, et procédé de fonctionnement du dispositif concentrateur et du système multi-dispositif
WO2023229279A1 (fr) Procédé de détermination de l'âge à l'aide d'un microbiome
EP3850509A1 (fr) Procédé et appareil d'authentification d'utilisateur
WO2023204488A1 (fr) Procédé et système de prédiction de la perte de cheveux selon le microbiome du cuir chevelu
WO2020256170A1 (fr) Dispostif de synthèse vocale faisant appel à l'intelligence artificielle, procédé de fonctionnement pour dispositif de synthèse vocale, et support d'enregistrement lisible par ordinateur
WO2023080275A1 (fr) Serveur de base de données d'applications de cadre d'apprentissage profond pour classifier le sexe et l'âge, et procédé associé
WO2021251579A1 (fr) Procédé de fourniture de données de base pour diagnostic, et système correspondant
WO2020230924A1 (fr) Appareil de synthèse de la parole utilisant l'intelligence artificielle, procédé de fonctionnement de l'appareil de synthèse de la parole, et support d'enregistrement lisible par ordinateur
WO2024005464A1 (fr) Procédé de clinique de données, programme informatique dans lequel un procédé de clinique de données est stocké et dispositif informatique qui effectue un procédé de clinique de données

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 23812053

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 24.03.2025)

122 Ep: pct application non-entry in european phase

Ref document number: 23812053

Country of ref document: EP

Kind code of ref document: A1