WO2022224465A1 - 血糖体質判定装置、血糖体質判定方法、および記録媒体 - Google Patents
血糖体質判定装置、血糖体質判定方法、および記録媒体 Download PDFInfo
- Publication number
- WO2022224465A1 WO2022224465A1 PCT/JP2021/030047 JP2021030047W WO2022224465A1 WO 2022224465 A1 WO2022224465 A1 WO 2022224465A1 JP 2021030047 W JP2021030047 W JP 2021030047W WO 2022224465 A1 WO2022224465 A1 WO 2022224465A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- question
- unit
- information
- type
- questions
- Prior art date
Links
- 239000008280 blood Substances 0.000 title claims abstract description 211
- 210000004369 blood Anatomy 0.000 title claims abstract description 211
- 238000000034 method Methods 0.000 title claims description 96
- 238000012545 processing Methods 0.000 claims abstract description 184
- 230000004044 response Effects 0.000 claims abstract description 36
- 238000010801 machine learning Methods 0.000 claims description 60
- WQZGKKKJIJFFOK-GASJEMHNSA-N Glucose Natural products OC[C@H]1OC(O)[C@H](O)[C@@H](O)[C@@H]1O WQZGKKKJIJFFOK-GASJEMHNSA-N 0.000 claims description 25
- 239000008103 glucose Substances 0.000 claims description 25
- 206010022489 Insulin Resistance Diseases 0.000 claims description 10
- 230000002641 glycemic effect Effects 0.000 claims description 10
- 230000003914 insulin secretion Effects 0.000 claims description 8
- 230000008569 process Effects 0.000 description 74
- 239000013598 vector Substances 0.000 description 71
- 238000007637 random forest analysis Methods 0.000 description 23
- 238000004422 calculation algorithm Methods 0.000 description 20
- 238000012360 testing method Methods 0.000 description 20
- 238000002790 cross-validation Methods 0.000 description 16
- 238000010586 diagram Methods 0.000 description 16
- 230000005540 biological transmission Effects 0.000 description 15
- 230000006870 function Effects 0.000 description 10
- 238000004891 communication Methods 0.000 description 9
- 238000007726 management method Methods 0.000 description 8
- 230000006872 improvement Effects 0.000 description 6
- 230000008859 change Effects 0.000 description 5
- 238000013135 deep learning Methods 0.000 description 5
- 230000036541 health Effects 0.000 description 5
- 238000003066 decision tree Methods 0.000 description 4
- 230000015654 memory Effects 0.000 description 4
- 241000699666 Mus <mouse, genus> Species 0.000 description 3
- 230000009471 action Effects 0.000 description 3
- 238000005259 measurement Methods 0.000 description 3
- 238000007639 printing Methods 0.000 description 3
- 230000003247 decreasing effect Effects 0.000 description 2
- 230000035622 drinking Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 206010018429 Glucose tolerance impaired Diseases 0.000 description 1
- 241000699670 Mus sp. Species 0.000 description 1
- 208000001280 Prediabetic State Diseases 0.000 description 1
- 208000032140 Sleepiness Diseases 0.000 description 1
- 206010041349 Somnolence Diseases 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000009534 blood test Methods 0.000 description 1
- 230000001680 brushing effect Effects 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 206010012601 diabetes mellitus Diseases 0.000 description 1
- 235000006694 eating habits Nutrition 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 201000009104 prediabetes syndrome Diseases 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 238000004579 scanning voltage microscopy Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000037321 sleepiness Effects 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
Images
Classifications
-
- 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
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/145—Measuring characteristics of blood in vivo, e.g. gas concentration, pH value; Measuring characteristics of body fluids or tissues, e.g. interstitial fluid, cerebral tissue
- A61B5/14532—Measuring characteristics of blood in vivo, e.g. gas concentration, pH value; Measuring characteristics of body fluids or tissues, e.g. interstitial fluid, cerebral tissue for measuring glucose, e.g. by tissue impedance measurement
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/20—Ensemble learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
-
- 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
- G16H10/00—ICT specially adapted for the handling or processing of patient-related medical or healthcare data
- G16H10/20—ICT specially adapted for the handling or processing of patient-related medical or healthcare data for electronic clinical trials or questionnaires
-
- 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
- G16H10/00—ICT specially adapted for the handling or processing of patient-related medical or healthcare data
- G16H10/60—ICT specially adapted for the handling or processing of patient-related medical or healthcare data for patient-specific data, e.g. for electronic patient records
-
- 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
- G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
-
- 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 blood sugar constitution determination device and the like for determining the type of blood sugar constitution.
- the health condition was predicted from the results of blood tests, etc. in health checkups. In addition, it was used to predict pre-diabetes and diabetes risk from general risk factors. More specifically, conventionally, for example, a method called a so-called sugar tolerance test has been used to monitor the user's health condition related to blood sugar. Specifically, for example, after the user ingests 75 g of glucose, every 30 minutes until 120 minutes, the blood glucose level of the user is measured by taking blood, testing the blood, and measuring the blood glucose level. is acquired, the time change is referred to, and the user's health condition regarding blood sugar determined by a doctor is observed.
- a blood sugar level information management device capable of presenting trigger information for causing a patient to continue taking actions to improve his/her own blood sugar level (see Patent Document 1).
- Such a device registers the medical information acquisition devices used by the patient, and obtains blood sugar level change information, which is information on changes in blood sugar levels within a predetermined period, based on biological information including blood sugar level information acquired from each medical information acquisition device. to obtain the answer information based on the lifestyle question information, which is a question about the lifestyle habits of the patient within the predetermined period, and based on the answer information, the degree of improvement of the lifestyle habit is determined.
- Improvement degree display information shown graphically is generated, and at least the biometric information change information, the lifestyle question information and its answer information, and the improvement degree display information are data that can be displayed on one sheet.
- the lifestyle question information includes only actions that contribute to improvement of blood sugar level
- the improvement degree display information is configured to indicate the degree of execution of the action that contributes to improvement of blood sugar level. It is a blood sugar level information management device.
- the blood sugar constitution determination device of the first invention is a storage unit for storing one or more question information including questions related to lifestyle habits, which is one or more question information for judging the type of blood sugar constitution, and one or more an output unit that outputs the question information, a reception unit that receives answer information for each of the one or more question information from the user, and one or more answer information received by the reception unit, to determine the type of blood sugar constitution of the user
- a blood sugar constitution determination device comprising a processing unit and an information output unit for outputting output information regarding a type of blood sugar constitution.
- the storage unit is one or more first questions for determining the presence or absence of risk related to blood sugar, and questions related to lifestyle habits.
- a first question storage unit that stores one or more first questions including The question information is the first question or the second question
- the output unit includes a first question output unit that outputs one or more first questions, and a first judgment unit a second question output unit that outputs one or more second questions when the user determines that there is a risk
- the reception unit receives a first answer to each of the one or more first questions from the user
- a receiving unit and a second answer receiving unit that receives second answers to each of the one or more second questions from the user
- the processing unit uses the one or more first answers received by the first answer receiving unit , a first determination unit that determines whether or not the user has a risk related to blood sugar; It is a blood glucose constitution determination device comprising.
- the second judgment unit also uses one or more first answers received by the first answer reception unit to determine the blood glucose level of the user. It is a blood glucose constitution determination device that determines the type of constitution.
- the type of blood sugar constitution of the user can be determined with high accuracy. More specifically, by determining the type of glycemic constitution of the user in the second stage, also using the first answer, for users at risk for glycemic constitution, using answers to fewer questions, As a result of being able to determine the type, the burden on the user can be reduced.
- the processing unit uses the one or more answer information received by the receiving unit and the learning device to predict the blood sugar constitution type by machine learning prediction processing It is a blood sugar constitution determination device that acquires
- the type of blood sugar constitution of the user can be determined with high accuracy.
- the machine learning further comprising a learning device storage unit in which the learning device acquired by the learning process of 1 is stored, and the second judgment unit uses the one or more second answers and the learning device received by the second answer receiving unit, It is a blood sugar constitution determination device that acquires the type of blood sugar constitution by prediction processing of machine learning.
- the type of blood sugar constitution of the user can be determined with high accuracy.
- the blood sugar constitution determination device of the sixth invention for any one of the first to fifth inventions, is associated with one or more type identifiers for identifying types of blood sugar constitution, and one or more advices further comprising an advice storage unit storing information, wherein the information output unit outputs output information including one or more pieces of advice information associated with the type identifier identifying the type determined by the second determination unit; It is a constitution determination device.
- one or more pieces of advice information associated with the type identifier that identifies the type determined by the second judgment unit is sent from the advice storage unit to the sixth invention.
- the apparatus for determining blood sugar constitution further comprising an output information configuration unit configured to acquire and configure output information having a type identifier and one or more pieces of advice information, wherein the information output unit outputs the output information configured by the output information configuration unit is.
- the one or more question information in the storage unit is a question acquired by the simple question sheet acquiring device, and the simple question sheet acquiring device is , are answers to N questions, and are obtained using answers to each of M questions, which are less than N, among the answers corresponding to any of the real type identifiers of 2 or 3 or more real type identifiers an estimated type acquisition unit that acquires an estimated type identifier;
- a question determining unit that determines questions that are corresponding M questions, and a simple question sheet output unit that outputs a simple question sheet that is information about a question set that is a set of M questions determined by the question determining unit.
- a blood glucose constitution determination device comprising:
- the one or more first questions in the first question storage unit are questions acquired by the simplified questionnaire acquisition device.
- the simplified questionnaire acquisition device is answers to N questions, and among N answers corresponding to actual risk presence/absence information specifying whether there is a risk or not, M answers less than N
- An estimation type acquisition unit that acquires estimated risk presence/absence information obtained using answers to each question, and the accuracy of the degree of matching between the estimated risk presence/absence information acquired by the estimation type acquisition unit and the actual risk presence/absence information are predetermined.
- Simple A blood glucose constitution determination device comprising a simple question form output unit for outputting a question form.
- one or more types of questions among the one or more second questions in the second question storage unit are simple questions It is a question acquired by the vote acquisition device, and the simple questionnaire acquisition device is an answer to N questions, and among the answers corresponding to any one of two or three or more real type identifiers,
- An estimated type acquisition unit that acquires an estimated type identifier obtained using answers to each of M questions, which is less than N, and an estimated type acquisition unit that has accuracy regarding the degree of matching between the estimated type identifier and the actual type identifier acquired by the estimated type acquisition unit.
- the type of blood sugar constitution is a combination of a type related to insulin sensitivity and a type related to insulin secretion. It is a blood glucose constitution determination device.
- the user's blood sugar constitution type can be easily known.
- the type of blood sugar constitution is a type related to the health condition related to blood sugar.
- the type of glycemic constitution is, for example, one of four types: a combination of types related to insulin sensitivity (either no resistance or tendency to resistance) and types related to insulin secretion (either normal or difficult to secrete). is.
- the type of blood sugar constitution is, for example, whether the blood sugar level tends to increase or not, or whether the blood sugar level tends to decrease.
- the number and types of blood sugar constitution types are not limited.
- the questionnaire preferably includes questions regarding matters that are not considered to directly affect blood sugar levels. It is preferable that the questionnaire include many questions that are easy to answer.
- Questions about lifestyle habits include, for example, questions about eating habits, questions about drinking or drinking habits, questions about exercise habits or lifestyle intensity, questions about sleep or sleepiness, questions about hygiene, and other general questions about lifestyle habits. is.
- General questions are, for example, questions about the frequency of brushing teeth, questions about the degree of use of stairs.
- a blood sugar constitution determination device that receives second answers to one or more additional second questions, acquires a blood sugar constitution type, and outputs the blood sugar constitution type.
- a blood sugar constitution determination apparatus will be described that allows a user who does not have a risk related to blood sugar to easily know the status of blood sugar by asking the user to answer in two steps.
- a blood sugar constitution determination device that acquires a blood sugar constitution type using one or more second answers and one or more first answers will be described.
- a description will be given of a blood sugar constitution determination device that acquires a blood sugar constitution type by prediction processing of machine learning using a learning model for obtaining a blood sugar constitution type.
- a description will be given of a blood sugar constitution determination device that acquires and outputs advice information using the type of blood sugar constitution.
- a description will be given of a blood sugar constitution determination device that obtains and outputs advice information using one or more answers in addition to the type of blood sugar constitution.
- a blood sugar constitution determination system including a blood sugar constitution determination device that receives an answer from a user terminal and transmits information such as the type of blood sugar constitution to the user terminal
- the blood sugar constitution determination device may be a stand-alone device.
- FIG. 1 is a conceptual diagram of a blood sugar constitution determination system A according to the present embodiment.
- a blood sugar constitution determination system A includes a learning device 1 , a blood sugar constitution determination device 2 , and one or more user terminals 3 .
- the learning device 1 is a device that acquires original information, which will be described later, using a type identifier that identifies the type of blood sugar constitution and answer information for each of two or more questions included in the questionnaire.
- the learning device 1 is, for example, a so-called personal computer, a tablet terminal, a smart phone, a server, or the like, and the type thereof does not matter.
- the server may be, for example, a cloud server or an ASP server, and any type of server may be used.
- the learning device 1 may be a stand-alone device or a server that receives instructions from the user terminal 3 and operates.
- the blood sugar constitution determination device 2 is, for example, a so-called server.
- the server may be, for example, a cloud server or an ASP server, and any type of server may be used.
- the blood sugar constitution determination device 2 may be a stand-alone device such as a so-called personal computer, tablet terminal, or smart phone.
- the user terminal 3 is a terminal used by a user who acquires the type of blood sugar constitution.
- the user terminal 3 is a so-called personal computer, a tablet terminal, a smart phone, or the like, and the type thereof does not matter.
- the learning device 1, the blood sugar constitution determination device 2, and the user terminal 3 may be able to communicate via the Internet, LAN, or the like.
- FIG. 2 is a block diagram of blood sugar constitution determination system A according to the present embodiment.
- FIG. 3 is a block diagram of the blood sugar constitution determination device 2. As shown in FIG.
- the learning device 1 includes a learning storage unit 11, a learning reception unit 12, a learning processing unit 13, and a learning output unit 14.
- the learning storage unit 11 includes an answer storage unit 111 and an actual type storage unit 112 .
- the blood sugar constitution determination device 2 includes a storage unit 21, a reception unit 22, a processing unit 23, and an output unit 24.
- the storage unit 21 includes a first question storage unit 211 , a second question storage unit 212 , an original information storage unit 213 and an advice storage unit 214 .
- the reception unit 22 includes a first reply reception unit 221 and a second reply reception unit 222 .
- the processing unit 23 includes a first determination unit 231 , a second determination unit 232 and an output information configuration unit 233 .
- the output unit 24 includes a first question output unit 241 , a second question output unit 242 and an information output unit 243 .
- the user terminal 3 includes a terminal storage unit 31, a terminal reception unit 32, a terminal processing unit 33, a terminal transmission unit 34, a terminal reception unit 35, and a terminal output unit 36.
- Various types of information are stored in the learning storage unit 11 that constitutes the learning device 1 .
- Various types of information are, for example, answer information and type.
- a type can also be called a type identifier.
- the type is, for example, one of "1", “2", “3”, and "4".
- Type "1” is, for example, information indicating a type of "normal” regarding insulin secretion and a type of "no resistance” regarding insulin sensitivity.
- Type “2” is, for example, information indicating a type of insulin secretion of "normal” and an insulin sensitivity of "tendency to resistance”.
- Type "3" is information indicating, for example, a type related to insulin secretion that is “difficult to appear” and a type related to insulin sensitivity that is “no resistance”.
- Type "4" is information indicating, for example, a type related to insulin secretion that is “difficult to appear” and a type related to insulin sensitivity that is “tendency to resist.”
- Type "1" is a normal type.
- Types "2", “3", and "4" are abnormal types.
- the answer may be called answer information. Also, the answer may be information specifying the user's answer to the question. An answer is an answer to a question. Answers are associated with real types. The actual type is the user's actual type. A real type may be called a real type identifier. Responses are associated with actual risk presence/absence information. The actual risk presence/absence information is information indicating whether or not the user has a risk related to blood sugar.
- the answer storage unit 111 stores, for example, answers associated with identifiers.
- the answer is the user's answer to each question on the questionnaire. Note that the user is the respondent.
- the answer is usually associated with the question identifier of each question on the question sheet.
- a question identifier is information for identifying a question, such as a question number or a question ID.
- the identifier is a user identifier, a day identifier, or a user identifier and a day identifier.
- a user identifier is information that identifies a user, and is, for example, an ID, e-mail address, telephone number, or ID of a terminal used by the user (eg, IP address, MAC address, terminal identifier, etc.).
- the answer may be associated with the user attribute value.
- User attribute values are user attribute values, such as age, height, weight, BMI, and gender.
- the answer is the answer to any of the questions of M (M is a natural number of 1 or more) selected from the questions of N (N is a natural number of 2 or more) by the simple question sheet acquisition device B, which will be described later. It is preferable that the information is indicative. In addition, it is "N>M".
- the answer storage unit 111 may store two or more first answers and two or more second answers.
- the first answer is an answer to the first question for determining the presence or absence of risk related to blood sugar.
- the first response corresponds to risk presence/absence information specifying whether there is a risk related to blood sugar.
- the second answer is an answer to the second question for judging the type of blood sugar constitution.
- the second answer is associated with the real type identifier when there is a risk.
- the first response and the second response are both associated with identifiers, for example.
- the real type storage unit 112 stores two or more real types.
- a real type may be called a real type identifier.
- the actual type is information identifying the actual type of glycemic constitution of the user who responded.
- a real type is usually associated with one or more answers.
- the actual type usually corresponds to a set of answers to each question on the questionnaire.
- a real type is associated with an identifier, for example.
- the real type is, for example, information acquired by a sugar tolerance test.
- the real type identifier is, for example, obtained by taking blood, testing the blood, and measuring the blood glucose level periodically (eg, every 30 minutes) for a period of 0 to 120 minutes after the user ingests 75 g of glucose.
- This is the type of blood sugar constitution determined by a doctor, a system, or the like by obtaining the time change of the user's blood sugar level and referring to the time change.
- the actual type may be information input manually.
- the actual type storage unit 112 may store two or more pieces of actual risk presence/absence information.
- the actual risk presence/absence information is usually associated with one or more answers.
- the actual risk presence/absence information is usually associated with a set of answers to each question on the questionnaire.
- the real risk presence/absence information is associated with an identifier, for example.
- the actual risk presence/absence information may be associated with an actual type.
- the actual risk presence/absence information is information specifying the presence/absence of the user's actual risk related to blood sugar.
- the actual risk presence/absence information is usually associated with one or more first responses.
- a real type is usually associated with one or more secondary answers.
- the real risk presence/absence information and the real type are associated with identifiers, for example.
- the learning reception unit 12 receives various instructions and information. Various instructions and information are, for example, learning instructions.
- a learning instruction is an instruction to acquire original information.
- Input means for various instructions and information may be anything, such as a touch panel, a keyboard, a mouse, or a menu screen. Note that the learning reception unit 12 may receive various instructions, information, and the like from the user terminal 3 .
- the original information is the information that becomes the basis for estimating the type from the answer information for each of two or more questions.
- the original information is, for example, a learning device to be described later.
- the original information is, for example, a correspondence table to be described later.
- the original information is usually acquired by the learning processing unit 13, which will be described later.
- the learning processing unit 13 acquires original information.
- the processing of the learning processing unit 13 is, for example, either (1) or (2) below.
- the learning processing unit 13 preferably acquires a learning device by performing the following processing (1).
- the original information is a learning device that determines the type of blood sugar constitution by one-time prediction processing
- the learning processing unit 13 acquires two or more sets from the learning storage unit 11.
- a set is a set of real types and two or more answers.
- An actual type in the set and a set of two or more answers in the set are associated with each other.
- An actual type in a set and a set of answers are associated with, for example, the same identifier. Note that such a set may be called teacher data.
- the learning processing unit 13 uses the two or more acquired pairs to perform learning processing according to a machine learning algorithm and acquires a learning device. More specifically, the learning processing unit 13, for example, provides the acquired two or more sets to a module that performs learning processing of machine learning, executes the module, and acquires a learner.
- the learning device may be called a classifier, a learning model, a model, or the like.
- this learner is a learner that is given to a module that performs prediction processing of machine learning together with a set of two or more answers to acquire a type.
- the machine learning learning processing module may be a random forest, a decision tree, deep learning, SVR, or the like, and any algorithm may be used.
- various machine learning functions such as TensorFlow library, fastText, tinySVM, random forest module of R language, and various existing libraries can be used.
- the learning processing unit 13 acquires two or more first pairs from the learning storage unit 11 .
- the first set here is a set of actual risk presence/absence information and one or more first answers.
- the actual risk presence/absence information indicates that there is a risk (for example, "1") and that there is no risk (for example, "0").
- the first answer here is the question of M 1 (M 1 is a natural number of 1 or more) selected from the questions of N 1 (N 1 is a natural number of 1 or more) by the simple question sheet acquisition device B, which will be described later. It is preferable that the information indicates an answer to any one of the questions. Note that N 1 >M 1 .
- the learning processing unit 13 uses the acquired two or more first sets to perform learning processing according to a machine learning algorithm, and acquires a first learner.
- the learning process may be the same process as the learning process described above.
- the first learning device is data used when performing prediction processing for receiving one or more first answers and acquiring estimated risk presence/absence information.
- the estimated risk presence/absence information is estimated risk presence/absence information.
- the learning processing unit 13 acquires two or more second sets from the learning storage unit 11 .
- the second set here is a set of actual types and one or more second answers.
- the first answer here is the question of M 2 (M 2 is a natural number of 1 or more) selected from the questions of N 2 (N 2 is a natural number of 1 or more) by the simple question sheet acquisition device B, which will be described later. It is preferable that the information indicates an answer to any one of the questions. Note that N 2 >M 2 .
- the learning processing unit 13 may acquire the second set and one or more first answers paired with the second set. Moreover, it is preferable that the one or more first answers acquired here are some of the first answers used when acquiring the first learner.
- the learning processing unit 13 uses the acquired second set of two or more to perform learning processing according to a machine learning algorithm, and acquires a second learner. Note that the learning processing unit 13 performs learning processing using a machine learning algorithm using two or more sets of the second set and one or more first answers paired with the second set. , it is preferable to obtain a second learner.
- the learning process may be the same process as the learning process described above.
- the second learning device is data used when performing prediction processing for receiving one or more second answers and obtaining an estimation type.
- the estimated type is the estimated type.
- the first algorithm of machine learning for obtaining the first learner and the second algorithm of machine learning for obtaining the second learner may be the same or different.
- the first algorithm is SVM and the second algorithm is random forest or deep learning.
- the original information is a correspondence table (2-1)
- the original information is a correspondence table for determining the type of blood sugar constitution
- the learning processing unit 13 acquires two or more sets from the learning storage unit 11.
- a set is a set of real types and two or more answers.
- the learning processing unit 13 acquires a vector whose elements are two or more answers for each set.
- the learning processing unit 13 acquires correspondence information indicating the correspondence between the vector and the actual type for each pair.
- the learning processing unit 13 constructs a correspondence table having correspondence information for each pair of two or more as a record.
- the correspondence information includes, for example, vectors and actual types.
- Correspondence information has, for example, a link to a vector and a link to an actual type.
- the learning processing unit 13 acquires two or more first pairs from the learning storage unit 11 .
- the first set here is a set of risk presence/absence information specifying the presence/absence of risk and two or more first responses.
- the learning processing unit 13 acquires a vector whose elements are two or more first answers for each set.
- the learning processing unit 13 acquires first correspondence information indicating the correspondence between the vector and the risk presence/absence information for each set.
- the learning processing unit 13 constructs a first correspondence table having, as a record, first correspondence information for each of the two or more first pairs. Note that the first correspondence information has, for example, the same data structure as the correspondence information described above.
- the learning processing unit 13 acquires two or more second sets from the learning storage unit 11 .
- the second set here is a set of actual types and two or more second answers.
- the learning processing unit 13 preferably acquires two or more sets of the second pair and one or more first answers paired with the second pair.
- the learning processing unit 13 acquires a vector whose elements are two or more second answers for each second set.
- the learning processing unit 13 acquires a vector whose elements are each of the two or more second answers and the one or more first answers of each of the two or more sets.
- the learning processing unit 13 acquires second correspondence information indicating the correspondence between the vector and the actual type for each second pair.
- the learning processing unit 13 constructs a second correspondence table having, as a record, the second correspondence information for each of the two or more second pairs.
- the second correspondence information has, for example, a vector and an actual type.
- the second correspondence information has, for example, a link to the vector and a link to the actual type.
- the learning output unit 14 outputs information.
- the learning output unit 14 outputs the original information acquired by the learning processing unit 13, for example.
- output usually means storage on a recording medium, but display on a display, projection using a projector, printing on a printer, sound output, transmission to an external device, or other processing device.
- the concept may also include delivery of processing results to other programs, etc.
- the learning output unit 14 accumulates the original information in the original information storage unit 213 of the blood sugar constitution determination device 2 .
- Various types of information are stored in the storage unit 21 that constitutes the blood sugar constitution determination device 2 .
- Various information includes, for example, two or more questions, one or two or more first questions described later, one or two or more second questions described later, one or two or more original information, and one or two or more user information.
- two or more questions may be referred to as a question sheet.
- one or more first questions may be referred to as the first question sheet.
- one or more second questions may be referred to as a second question sheet.
- the question is the question information for estimating the type. It is preferable that the two or more questions include questions about lifestyle habits. Also, the question may be the first question to be described later or the second question to be described later.
- the first question storage unit 211 stores one or more first questions.
- the first question is a question for judging whether or not there is a risk related to blood sugar.
- the one or more first questions preferably include one or more questions regarding lifestyle habits.
- the second question storage unit 212 stores one or more second questions.
- the second question is a question for judging the type of blood sugar constitution.
- the one or more second questions preferably include one or more questions regarding lifestyle habits.
- the type of blood sugar constitution is a combination of the type related to insulin sensitivity and the type related to insulin secretion.
- the type for insulin sensitivity is, for example, either “no resistance” or “prone to resistance”.
- the type for insulin sensitivity is, for example, either "normal” or “poor”.
- the original information storage unit 213 stores one or more pieces of original information.
- the original information is usually information acquired by the learning device 1 .
- the source information is, for example, the above-described learning device, the above-described first learning device, the above-described second learning device, the above-described correspondence table, the above-described first correspondence table, or the above-described second correspondence table.
- the advice storage unit 214 stores one or more pieces of advice information in association with one or more type identifiers that identify types of blood sugar constitution.
- the content of the advice information does not matter.
- One or more pieces of advice information may be stored in the advice storage unit 214 in association with the risk presence/absence information.
- the reception unit 22 receives various instructions and information. Various instructions and information are, for example, an answer, a first answer, and a second answer. The reception unit 22 receives, for example, answers to two or more questions.
- Acceptance usually means reception from the user terminal 3, but acceptance of information input from input devices such as keyboards, mice, and touch panels, and reading from recording media such as optical discs, magnetic discs, and semiconductor memories. It may also be a concept including acceptance of received information.
- the first answer accepting unit 221 accepts a first answer to each of one or more first questions from the user.
- the first reply receiving unit 221 receives one or more first replies from the user terminal 3, for example.
- the second answer accepting unit 222 accepts second answers to one or more second questions from the user.
- the processing unit 23 performs various types of processing.
- the processing unit 23 is processing performed by the first determination unit 231, the second determination unit 232, and the output information configuration unit 233, for example.
- the processing unit 23 or a determination unit acquires the type using, for example, one or more answers received by the reception unit 22.
- the processing unit 23 or a determination unit performs machine learning prediction processing using, for example, one or more answers received by the reception unit 22 and a learning device, and acquires a type. More specifically, the processing unit 23 or the determination unit (not shown) provides, for example, one or more answers received by the reception unit 22 and a learning device to a module that performs machine learning prediction processing, and executes the module. and get the type.
- the prediction processing module of machine learning may be random forest, decision tree, deep learning, SVM, or the like, and any algorithm may be used.
- various machine learning functions such as TensorFlow library, fastText, tinySVM, R language random forest module, and various existing libraries can be used.
- the learning device is a learning device acquired by the learning device 1 .
- the processing unit 23 or a determination unit acquires the type, for example, using one or more answers received by the reception unit 22 and the correspondence table. More specifically, the processing unit 23 or the determination unit (not shown) constructs a vector whose elements are one or more responses received by the reception unit 22, for example. Next, the processing unit 23 or the determination unit (not shown) searches the correspondence table for the vector that most closely approximates the vector, and acquires the type paired with the vector that most closely approximates the vector from the correspondence table.
- the first determination unit 231 uses one or more first responses received by the first response reception unit 221 to determine whether or not the user has a risk related to blood sugar.
- the first judgment unit 231 uses one or more first answers received by the first answer reception unit 221 and the first learning device to perform machine learning prediction processing and acquire estimated risk presence/absence information. More specifically, the first determination unit 231 provides, for example, one or more first answers accepted by the first answer acceptance unit 221 and the first learner to a module that performs machine learning prediction processing, to obtain estimated risk information.
- the first judgment unit 231 constructs a vector whose elements are, for example, one or more first answers received by the first reply reception unit 221 .
- the first determination unit 231 searches the first correspondence table for a vector that is most similar to the vector, and acquires the risk presence/absence information paired with the closest vector from the first correspondence table. This risk presence/absence information is estimated risk presence/absence information.
- the second determination unit 232 uses one or more second answers accepted by the second answer acceptance unit 222 to determine the type of blood sugar constitution of the user.
- the second judgment unit 232 uses the one or more second answers received by the second answer reception unit 222 and the second learning device to perform machine learning prediction processing and acquires the estimated type. More specifically, the second determination unit 232 provides, for example, one or more second answers accepted by the second answer acceptance unit 222 and the second learning device to a module that performs machine learning prediction processing, to get the inferred type.
- the second judgment unit 232 constructs a vector whose elements are, for example, one or more second answers received by the second reply reception unit 222 .
- the second determination unit 232 searches the second correspondence table for the vector that most closely approximates the vector, and acquires the type paired with the closest vector from the second correspondence table. Note that this type is an estimation type.
- the second judgment unit 232 also uses one or more first answers received by the first reply reception unit 221 to determine the type of blood sugar constitution of the user.
- the one or more first responses are preferably part of the first responses used when acquiring the risk presence/absence information.
- the second judgment unit 232 uses, for example, one or more second answers accepted by the second answer acceptance unit 222, one or more first answers accepted by the first answer acceptance unit 221, and a second learning device, Perform machine learning prediction processing and get the inference type. More specifically, the second determination unit 232 determines, for example, one or more second answers accepted by the second answer acceptance unit 222, one or more first answers accepted by the first answer acceptance unit 221, and the second learning device is given to a module that performs machine learning prediction processing, the module is executed, and an estimation type is obtained.
- the second determination unit 232 constructs a vector whose elements are, for example, one or more second answers accepted by the second answer acceptance unit 222 and one or more first answers accepted by the first answer acceptance unit 221. do.
- the second determination unit 232 searches the second correspondence table for the vector that most closely approximates the vector, and acquires the type paired with the closest vector from the second correspondence table. Note that this type is an estimation type.
- the output information configuration unit 233 acquires from the advice storage unit 214 one or more pieces of advice information associated with the type identifier that identifies the type determined by the second determination unit 232, and stores the type identifier and one or more pieces of advice information. Configure the output information to have.
- the output unit 24 outputs various information.
- Various types of information are, for example, questions, first questions, second questions, and output information.
- the output is, for example, transmission to the user terminal 3, display on a display, projection using a projector, printing on a printer, sound output, storage on a recording medium, other processing devices or other
- the concept may include delivery of the processing result to a program or the like.
- the first question output unit 241 outputs one or more first questions from the first question storage unit 211.
- the second question output unit 242 outputs one or more second questions from the second question storage unit 212 when the first determination unit 231 determines that there is a risk.
- the information output unit 243 outputs output information regarding the type of blood sugar constitution.
- the information output unit 243 outputs output information including, for example, one or more pieces of advice information associated with the estimation type determined by the second determination unit 232 .
- the information output unit 243 outputs the output information configured by the output information configuration unit 233, for example.
- Various types of information are stored in the terminal storage unit 31 that constitutes the user terminal 3 .
- Various types of information are, for example, a user identifier and one or more user attribute values.
- the terminal reception unit 32 receives various instructions and information.
- Various instructions and information are, for example, one or more answers and a question sheet output instruction.
- the answer may be the first answer or the second answer.
- the question sheet output instruction is an instruction to output the question sheet.
- the question sheet is preferably a simple question sheet created by a simple question sheet acquisition device B, which will be described later.
- Various instructions and information input means can be anything, such as a touch panel, keyboard, mouse, or menu screen.
- the terminal processing unit 33 performs various types of processing.
- Various types of processing are, for example, converting instructions, information, etc. received by the terminal reception unit 32 into a data structure of instructions, information, etc. to be transmitted.
- Various types of processing are, for example, converting the information received by the terminal reception unit 35 into a data structure for output.
- the terminal transmission unit 34 transmits various instructions, information, etc. to the blood glucose constitution determination device 2 .
- the terminal transmission unit 34 transmits instructions, information, and the like configured by the terminal processing unit 33 to the blood glucose constitution determination device 2 .
- the terminal reception unit 35 receives various information and the like.
- the terminal receiving unit 35 receives, for example, a questionnaire, estimated risk presence/absence information, estimated type, advice information, or output information from the blood sugar constitution determination device 2 .
- the terminal output unit 36 outputs various information and the like.
- the terminal output unit 36 outputs, for example, a questionnaire, estimated risk presence/absence information, estimated type, advice information, or output information.
- Learning storage unit 11, answer storage unit 111, actual type storage unit 112, storage unit 21, first question storage unit 211, second question storage unit 212, original information storage unit 213, advice storage unit 214, and terminal storage unit 31 is preferably a non-volatile recording medium, but can also be realized with a volatile recording medium.
- the process by which information is stored in the learning storage unit 11 or the like does not matter.
- information may be stored in the learning storage unit 11 or the like via a recording medium, or information transmitted via a communication line or the like may be stored in the learning storage unit 11 or the like.
- information input via an input device may be stored in the learning storage unit 11 or the like.
- the learning reception unit 12 and the terminal reception unit 32 can be realized by device drivers for input means such as touch panels and keyboards, control software for menu screens, and the like.
- the learning receiving unit 12 may be realized by wireless or wired communication means.
- the learning processing unit 13, the learning output unit 14, the processing unit 23, the first determination unit 231, the second determination unit 232, the output information configuration unit 233, and the terminal processing unit 33 can usually be implemented by a processor, memory, or the like.
- the processing procedure of the learning processing unit 13 and the like is normally realized by software, and the software is recorded in a recording medium such as a ROM. However, it may be realized by hardware (dedicated circuit).
- the processor is, for example, a CPU, MPU, GPU, or the like, and any type of processor does not matter.
- learning output unit 14 may be realized by wireless or wired communication means.
- the receiving unit 22, the first reply receiving unit 221, the second reply receiving unit 222, and the terminal receiving unit 35 are usually realized by wireless or wired communication means, but may be realized by means of receiving broadcast. .
- the output unit 24, the first question output unit 241, the second question output unit 242, the information output unit 243, and the terminal transmission unit 34 are usually implemented by wireless or wired communication means, but are implemented by broadcasting means. Also good.
- the terminal reception unit 32 can be realized by device drivers for input means such as touch panels and keyboards, control software for menu screens, and the like.
- the terminal output unit 36 may or may not include output devices such as displays and speakers.
- the terminal output unit 36 can be realized by output device driver software, or by output device driver software and an output device.
- a first example of operation is an example of an operation that constitutes two learners, a first learner and a second learner.
- Step S401 The learning processing unit 13 substitutes 1 for the counter i.
- Step S402 The learning processing unit 13 determines whether or not the i-th identifier exists. If the i-th identifier exists, go to step S403, and if the i-th identifier does not exist, go to step S406. Note that the identifier is associated with one or more first answers and one or more second answers.
- Step S403 The learning processing unit 13 acquires from the learning storage unit 11 a set of first answers paired with the i-th identifier and risk presence/absence information paired with the set of first answers.
- the learning processing unit 13 may determine and acquire the risk presence/absence information from the type.
- the type identifier “1” is the risk presence/absence information “absent “0””
- the type identifiers “2”, “3” or “4” is the risk presence information “present “1””.
- Step S404 The learning processing unit 13 constructs the i-th teacher data using the set of i-th first answers and the risk presence/absence information acquired in step S403, and accumulates it in a buffer (not shown).
- Step S405 The learning processing unit 13 increments the counter i by 1. Return to step S402.
- Step S406 The learning processing unit 13 acquires the two or more teacher data accumulated in step S404 from a buffer (not shown), and uses the two or more teacher data to perform prediction processing of machine learning. to get
- Step S407 The learning processing unit 13 accumulates the first learner acquired in step S406 in the learning storage unit 11.
- Step S408 The learning processing unit 13 substitutes 1 for the counter j.
- Step S409 The learning processing unit 13 determines whether or not the j-th identifier exists. If the j-th identifier exists, go to step S410; if the j-th identifier does not exist, go to step S415. Note that the identifier is associated with one or more first answers and one or more second answers.
- Step S410 The learning processing unit 13 acquires a set of second answers paired with the j-th identifier and an actual type paired with the set of second answers.
- a set of second answers exists in the answer storage unit 111 .
- the real type exists in the real type storage unit 112 .
- Step S411 The learning processing unit 13 determines whether or not to use one or more first answers to configure the second learning device. If the first answer is used, go to step S412; if not, go to step S413. It should be noted that whether or not to use the first reply is usually decided in advance.
- the learning processing unit 13 stores the one or more first answers that are paired with the j-th identifier and used to configure the second learning device in the answer storage unit 111. Get from One or more first answers used to configure the second learner are usually predetermined.
- Step S413 The learning processing unit 13 uses the set of j-th second answers obtained in step S410 and the actual type to compose j-th teacher data and store it in a buffer (not shown), or step S410. Using the set of j-th second answers obtained in step S412, the one or more first answers obtained in step S412, and the actual type, j-th teacher data is constructed and stored in a buffer (not shown). Note that the learning processing unit 13, for example, constructs the j-th teacher data, which is a vector whose elements are each answer and the actual type, and accumulates it in a buffer (not shown).
- Step S414 The learning processing unit 13 increments the counter j by 1. Return to step S409.
- Step S415) The learning processing unit 13 acquires the two or more teacher data accumulated in step S413 from a buffer (not shown), and uses the two or more teacher data to perform prediction processing of machine learning. to get
- Step S416) The learning processing unit 13 accumulates the second learner acquired in step S415 in the learning storage unit 11. End the process.
- a second operation example is an example of an operation that constitutes one learning device.
- One such learner is a learner for estimating types.
- Step S501 The learning processing unit 13 substitutes 1 for the counter i.
- Step S502 The learning processing unit 13 determines whether or not the i-th identifier exists. If the i-th identifier exists, go to step S403, and if the i-th identifier does not exist, go to step S406. Note that the identifier is associated with a set of one or more pieces of answer information.
- Step S503 The learning processing unit 13 acquires a set of answers paired with the i-th identifier and an actual type paired with the set of answers.
- Step S504 The learning processing unit 13 constructs the i-th teacher data using the set of i-th answers and the actual type obtained in step S503, and stores it in a buffer (not shown).
- Step S505 The learning processing unit 13 increments the counter i by 1. Return to step S402.
- Step S506 The learning processing unit 13 acquires the two or more teacher data accumulated in step S504 from a buffer (not shown), and uses the two or more teacher data to perform machine learning prediction processing to acquire a learner. do.
- Step S507 The learning processing unit 13 accumulates the learning device acquired in step S506 in the learning storage unit 11. End the process.
- Step S601 The reception unit 22 determines whether or not a question sheet output instruction has been received. If the question sheet output instruction has been received, the process proceeds to step S602, and if the question sheet output instruction has not been received, the process returns to step S601. Here, for example, the receiving unit 22 determines whether or not a question sheet output instruction has been received from the user terminal 3 .
- Step S602 The processing unit 23 acquires the first question sheet from the first question storage unit 211.
- Step S603 The first question output unit 241 outputs the first question sheet acquired in step S602. Here, the first question output unit 241 transmits the first question sheet to the user terminal 3, for example.
- Step S604 The first reply receiving unit 221 determines whether or not a set of one or more first replies has been received. If the first answer set has been accepted, the process goes to step S605, and if the first answer set has not been accepted, the process returns to step S604. Here, the first reply receiving unit 221 determines whether or not a set of one or more first replies has been received from the user terminal 3, for example.
- Step S605 The first determination unit 231 uses the set of one or more first answers accepted in step S604 to construct a first vector, which is a vector having each first answer as an element.
- Step S606 The first judgment unit 231 acquires the first learner from the original information storage unit 213.
- Step S607 The first determination unit 231 uses the first vector acquired in step S605 and the first learner acquired in step S606 to perform a first prediction process, which is a machine learning prediction process, and estimate Acquire risk presence/absence information.
- a first prediction process which is a machine learning prediction process, and estimate Acquire risk presence/absence information.
- Step S608 The first determination unit 231 determines whether the estimated risk presence/absence information acquired in step S607 is "present” or "absent”. If the risk is "present”, go to step S609, and if the risk is "absent”, go to step S620.
- Step S609 The processing unit 23 acquires the second question sheet from the second question storage unit 212.
- Step S610 The second question output unit 242 outputs the second question sheet acquired in step S609.
- the second question output unit 242 transmits the second question sheet to the user terminal 3, for example.
- Step S611 The second reply receiving unit 222 determines whether or not a set of one or more second replies has been received. If the set of second answers has been accepted, the process goes to step S612, and if the set of second answers has not been accepted, the process returns to step S611. In addition, here, the second reply receiving unit 222 determines whether or not a set of one or more second replies has been received from the user terminal 3, for example.
- Step S612 The second judgment unit 232 judges whether or not to use one or more first answers in the second prediction process. If the first answer is used, go to step S613; if not, go to step S614.
- Step S613 The second judgment unit 232 acquires one or more first answers to be used in the second prediction process from among the first answers received in step S604.
- the second determination unit 232 constructs a second vector.
- the second vector is a vector whose elements are each of the one or more second answers, or a vector whose elements are each of the one or more second answers and each of the one or more first answers obtained in step S613.
- Step S615 The second determination unit 232 acquires the second learning device from the original information storage unit 213.
- Step S616 The second judgment unit 232 uses the second vector acquired in step S614 and the second learner acquired in step S615 to perform second prediction processing, which is a prediction processing of machine learning, and estimate Get the type identifier.
- Step S617 The output information configuration unit 233 acquires from the advice storage unit 214 advice information paired with the estimated type identifier acquired in step S616.
- Step S618 The output information configuration unit 233 configures output information using the estimated type identifier acquired in step S616 and the advice information acquired in step S617.
- Step S619 The information output unit 243 outputs the output information configured in step S618. End the process.
- the information output unit 243 transmits output information to the user terminal 3, for example.
- Step S620 The first determination unit 231 substitutes information indicating "no risk” (here, "1") for the variable "type”.
- Step S621 The output information configuration unit 233 configures output information using the estimated type identifier "1" acquired in step S620.
- Step S622 The information output unit 243 outputs the output information configured in step S621. End the process.
- the information output unit 243 transmits output information to the user terminal 3, for example.
- Step S701 The reception unit 22 determines whether or not a question sheet output instruction has been received. If the question sheet output instruction is received, the process proceeds to step S702, and if the question sheet output instruction is not received, the process returns to step S701. Here, for example, the receiving unit 22 determines whether or not a question sheet output instruction has been received from the user terminal 3 .
- Step S702 The processing unit 23 acquires the question sheet from the storage unit 21.
- a question sheet is a set of questions.
- Step S703 The output unit 24 outputs the question sheet acquired in step S702. In addition, the output part 24 transmits a question sheet to the user terminal 3 here, for example.
- Step S704 The reception unit 22 determines whether or not a set of one or more pieces of answer information has been received. If the set of answers has been received, the process goes to step S705, and if the set of answers has not been received, the process returns to step S704. Here, for example, the reception unit 22 determines whether or not a set of one or more answers has been received from the user terminal 3 .
- Step S705 The processing unit 23 (can also be a determination unit (not shown)) uses the set of one or more answers accepted in step S704 to construct a vector having each answer as an element.
- Step S706 The processing unit 23 (can also be a determination unit not shown) acquires a learning device from the original information storage unit 213.
- Step S707 The processing unit 23 (can also be a determination unit not shown) performs machine learning prediction processing using the vector acquired in step S705 and the learning device acquired in step S706, and acquires an estimation type. .
- Step S708 The output information configuration unit 233 acquires from the advice storage unit 214 advice information paired with the estimated type acquired in step S707.
- Step S709 The output information configuration unit 233 configures output information using the estimated type acquired in step S707 and the advice information acquired in step S708.
- Step S710 The information output unit 243 outputs the output information configured in step S709. End the process.
- the information output unit 243 transmits output information to the user terminal 3, for example.
- Step S801 The terminal reception unit 32 determines whether or not a question sheet output instruction has been received. If the question sheet output instruction has been received, the process proceeds to step S802, and if the question sheet output instruction has not been received, the process returns to step S801.
- Step S802 The terminal processing unit 33 composes a question sheet output instruction to be sent.
- the terminal transmission unit 34 transmits the configured question sheet output instruction to the blood glucose constitution determination device 2 .
- Step S803 The terminal reception unit 35 determines whether or not the first question sheet has been received. When the first question sheet has been received, the process goes to step S804, and when it has not been received, the process returns to step S803.
- Step S804 The terminal processing unit 33 composes the first question sheet to be output from the first question sheet received in step S803.
- the terminal output unit 36 outputs the first question sheet.
- Step S805 The terminal reception unit 32 determines whether or not a set of first answers has been received from the user. If the set of first answers has been accepted, the process goes to step S806, and if the set of first answers has not been accepted, the process returns to step S805.
- Step S806 The terminal processing unit 33 constructs a set of first responses to be transmitted.
- the terminal transmission unit 34 transmits the configured set of first answers to the blood glucose constitution determination device 2 .
- Step S807 The terminal reception unit 35 determines whether information has been received from the blood sugar constitution determination device 2 or not. If the information has been received, the process goes to step S808, and if the information has not been received, the process returns to step S807.
- Step S808 The terminal processing unit 33 determines whether the information received in step S807 is output information. If it is output information, go to step S809, otherwise go to step S810. The information received in step S807 is the output information or the second question sheet.
- Step S809 The terminal output unit 36 outputs the output information received in step S807. End the process.
- Step S810 The terminal output unit 36 outputs the second question sheet received in step S807.
- Step S811 The terminal reception unit 32 determines whether or not the set of second answers has been received from the user. If the set of second answers has been accepted, the process goes to step S812, and if the set of second answers has not been accepted, the process returns to step S811.
- Step S812 The terminal processing unit 33 constructs a set of second responses to be transmitted.
- the terminal transmission unit 34 transmits the configured set of second answers to the blood sugar constitution determination device 2 .
- Step S813 The terminal reception unit 35 determines whether or not the output information has been received from the blood sugar constitution determination device 2. If the output information has been received, the process goes to step S814, and if the output information has not been received, the process returns to step S813.
- Step S814 The terminal output unit 36 outputs the received output information. End the process.
- the user terminal 3 accepts answers to each of the two or more questions included in the questionnaire, sends a set of the answers to the blood sugar constitution determination device 2, and responds to the transmission, including the type
- the output information may be received from the blood glucose constitution determination device 2 .
- a user information management table is a table for managing one or more pieces of user information.
- the user information management table manages one or more records having "ID”, "user identifier” and "user attribute value”.
- the first question sheet shown in FIG. 10 is stored in the first question storage unit 211, for example.
- the first question sheet has fifteen first questions. It is also assumed that the first question sheet is a simple question sheet acquired by a simple question sheet acquiring apparatus B, which will be described later. Also, the 15 first questions are questions for one or more first answers used to acquire the first learner.
- the second question sheet shown in FIG. 11 is stored in the second question storage unit 212, for example.
- the second question sheet has eight second questions. It is also assumed that the second question sheet is a simple question sheet acquired by a simple question sheet acquisition device B, which will be described later. Also, the second question of 8 is a question for one or more second answers used to acquire the second learner.
- the original information storage unit 213 stores a first learning device and a second learning device.
- the first learning device here is a model for estimating risk presence/absence information using one or more user attribute values and one or more first answers.
- the second learner is here a model for estimating at-risk person types using one or more user attribute values, one or more first responses, and one or more second responses.
- the learning processing unit 13 of the learning device 1 performs learning processing of machine learning using two or more teacher data having one or more user attribute values and one or more first answers, and selects the first learner. It is assumed that the information is obtained and accumulated in the original information storage unit 213 of the blood glucose constitution determination device 2 .
- the learning processing unit 13 of the learning device 1 performs learning processing of machine learning using two or more teacher data having one or more user attribute values, one or more first answers, and one or more second answers.
- the second learning device is acquired and stored in the original information storage unit 213 of the blood sugar constitution determination device 2.
- the one or more first answers used to acquire the second learner are partial answers of the one or more first answers used to acquire the first learner.
- the first question corresponding to each of the one or more first answers and the second question corresponding to each of the one or more second answers are both simple questions described later from the same large number of questions (for example, 350 questions) Assume that the question is determined by the vote acquisition device B.
- advice information is stored in the advice storage unit 214 in association with the type.
- the terminal reception unit 32 receives a question sheet output instruction.
- the terminal processing unit 33 composes a question sheet output instruction to be transmitted.
- the terminal transmission unit 34 transmits the configured question sheet output instruction to the blood glucose constitution determination device 2 .
- the reception unit 22 of the blood sugar constitution determination device 2 receives the question sheet output instruction.
- the processing unit 23 acquires the first question sheet having fifteen first questions (see FIG. 10) from the first question storage unit 211 .
- the first question output unit 241 transmits the first question sheet to the user terminal 3 .
- the terminal receiving section 35 of the user terminal 3 receives the first question sheet.
- the terminal processing unit 33 configures the first question sheet to be output from the received first question sheet.
- the terminal output unit 36 outputs the first question sheet.
- FIG. 12(b) An example of such an output is shown in FIG. 12(b). According to FIG. 12(b), here, the first question is output one by one, and the user answers one by one.
- the terminal processing unit 33 composes a set of first answers to be sent.
- the terminal transmission unit 34 transmits the configured set of first answers and the user identifier “U001” to the blood glucose constitution determination device 2 .
- the first response receiving unit 221 of the blood sugar constitution determination device 2 receives a set of one or more first responses and the user identifier "U001".
- the first determination unit 231 uses the received set of one or more first answers to construct a first vector, which is a vector having each first answer as an element. That is, the first determination unit 231 acquires the user attribute values (here, age, weight, BMI, and height) paired with the user identifier "U001" from the user information management table of FIG. Also, the first determination unit 231 acquires a set of 15 received first responses. Then, the first determination unit 231 determines a first vector (for example, (56, 78, 27.0, 170, first response 1, first response 2, . . . constitute the first answer 15)).
- a first vector for example, (56, 78, 27.0, 170, first response 1, first response 2, . . . constitute the first answer 15).
- the first judgment unit 231 acquires the first learning device from the original information storage unit 213. Then, the first determination unit 231 uses the first vector and the acquired first learner to perform the first prediction process, which is a machine learning prediction process, and acquires risk presence/absence information. Here, it is assumed that the first judging unit 231 has obtained risk presence/absence information indicating “risk present “1””.
- the processing unit 23 acquires from the second question storage unit 212 the screen information including the second question sheet corresponding to "1" with risk.
- the second question output unit 242 transmits screen information including the second question sheet to the user terminal 3.
- the terminal reception unit 35 receives information from the blood glucose constitution determination device 2 .
- the terminal output unit 36 outputs the received information. An example of such an output is shown in FIG. 12(c).
- the terminal reception unit 32 receives a set of second answers from the user.
- the terminal processing unit 33 composes a set of second responses to be transmitted.
- the terminal transmission unit 34 transmits the configured set of second answers to the blood sugar constitution determination device 2 .
- the second response reception unit 222 of the blood sugar constitution determination device 2 receives a set of eight second responses.
- the second determination unit 232 acquires the user attribute values (here, age, weight, BMI, height) paired with the user identifier "U001" from the user information management table of FIG.
- the second determination unit 232 also acquires a set of eight second responses received.
- the second determination unit 232 determines a second vector (for example, (56, 78, 27.0 , 170, first answer 1, first answer 2, . . . , second answer 1, second answer 2, .
- the second judgment unit 232 acquires the second learning device from the original information storage unit 213.
- the second determination unit 232 performs a second prediction process, which is a prediction process of machine learning, using the obtained second vector and the obtained second learner, and obtains the estimation type "2".
- the output information configuration unit 233 acquires advice information paired with the type "2" from the advice storage unit 214.
- advice information is, for example, the information on the screen of FIG. 12(f).
- the output information configuration unit 233 configures output information using the acquired estimation type "2" and the acquired advice information.
- the information output unit 243 transmits the output information to the user terminal 3.
- the terminal reception unit 35 of the user terminal 3 receives the output information from the blood sugar constitution determination device 2.
- the terminal output unit 36 outputs the received output information. Such output examples are shown in FIGS. 12(e) and 12(f).
- the user can easily know the type of blood sugar constitution of the user by answering simple questions.
- the user who is at risk for blood sugar can use answers to few questions.
- the burden on the user can be reduced.
- the processing in the present embodiment may be realized by software. Then, this software may be distributed by software download or the like. Also, this software may be recorded on a recording medium such as a CD-ROM and distributed. Note that this also applies to other embodiments in this specification.
- the software that implements the blood sugar constitution determination device 2 in the present embodiment is the following program.
- the program comprises one or more pieces of question information for judging the type of blood sugar constitution, and a computer capable of accessing a storage unit storing one or more pieces of question information including questions about lifestyle habits.
- FIG. 13 is a block diagram of simple question sheet acquisition device B in this embodiment.
- the simplified question sheet acquisition device B includes a storage unit 4 , a reception unit 5 , a processing unit 6 and an output unit 7 .
- the storage unit 4 includes a question sheet storage unit 41 , an answer storage unit 42 and an actual type storage unit 43 .
- the reception unit 5 includes an answer reception unit 51 .
- the processing unit 6 includes an actual type acquisition unit 61 , an inappropriateness determination unit 62 , an estimated type acquisition unit 63 and a question determination unit 64 .
- the estimation type acquisition unit 63 includes contribution degree acquisition means 631 , candidate acquisition means 632 , and estimation type acquisition means 633 .
- the output unit 7 includes a simple question sheet output unit 71 .
- Various types of information are stored in the storage unit 4.
- Various types of information are, for example, a question sheet to be described later, an answer to be described later, an actual type to be described later, an inappropriate condition to be described later, and an approximate condition to be described later.
- a question sheet is stored in the question sheet storage unit 41 .
- a question sheet is a set of N (N is a natural number) questions.
- a question sheet is, for example, a set of questions including M 1 (M 1 is a natural number, M 1 ⁇ N) questions used when acquiring an estimated type.
- a question sheet is, for example, a set of questions including M 2 (M 2 is a natural number, M 2 ⁇ N) questions used when acquiring estimated risk presence/absence information.
- a question sheet is, for example, a set of questions for estimating the type of user's glycemic constitution.
- the questionnaire is, for example, a set of questions for determining whether or not the user is at risk for blood sugar. However, the contents of the questionnaire and the number of questions are not considered.
- the question may also be referred to as a question, a question, or the like. It can be said that the questionnaire is a so-called questionnaire.
- the questionnaire consists of, for example, 350 questions.
- the answer is stored in the answer storage unit 42.
- the answer storage unit 42 stores answers associated with identifiers.
- the answer is the user's answer to each question on the questionnaire. Note that the user is the respondent.
- the answer corresponds to the question identifier of each question on the question sheet.
- a question identifier is information for identifying a question, such as a question number or a question ID.
- An identifier is information that identifies a set of responses, such as a user identifier, a day identifier, or a user identifier and a day identifier.
- a user identifier is information that identifies a user, and is, for example, an ID, e-mail address, telephone number, or ID of a terminal used by the user (eg, IP address, MAC address, terminal identifier, etc.).
- the answer may be associated with the user attribute value.
- User attribute values are user attribute values, such as gender, age, and occupation.
- the real type storage unit 43 stores two or more real types.
- a real type is usually associated with an identifier.
- a real type may be associated with an application that uses the real type. That is, the actual type storage unit 43 may store the actual type for learning and the actual type for testing separately. The real type storage unit 43 may store a real type for learning, a real type for verification, and a real type for testing, separately.
- the reception unit 5 receives various instructions and information.
- Various instructions and information are, for example, answers and simple question sheet output instructions.
- the simple question sheet output instruction is an instruction to output the simple question sheet.
- Various instructions and information input means can be anything, such as a touch panel, keyboard, mouse, or menu screen.
- the answer acceptance unit 51 accepts answers. Answers are answers to each question on the questionnaire. Answers are typically associated with user identifiers.
- the answer accepting unit 51 accepts a set of answers to each question on the questionnaire, associates the set of answers with the user identifier, and stores them in the answer storage unit 42 .
- the processing unit 6 performs various types of processing. Various types of processing are performed by, for example, the actual type acquiring unit 61, the inappropriateness determining unit 62, the estimated type acquiring unit 63, the question determining unit 64, the contribution degree acquiring unit 631, the candidate acquiring unit 632, and the estimated type acquiring unit 633. be.
- the actual type acquisition unit 61 acquires the actual type identifier from the actual type storage unit 43.
- the actual type acquisition unit 61 acquires from the actual type storage unit 43 the actual type paired with the set of answers to be processed.
- the actual type paired with the set of answers to be processed is, for example, the set of answers associated with the same identifier and the actual type.
- the inappropriate determination unit 62 determines whether or not the answers to the N questions satisfy predetermined inappropriate conditions.
- An inappropriate condition is, for example, different answers to the same two or more questions.
- the inadequate condition is, for example, that the proportion of responses equal to or greater than a threshold (eg, 95% or greater) or that the number of responses equal to or greater than the threshold are the same (eg, that all responses are the same). It is preferable that the estimated type acquisition unit 63 and the question determination unit 64 not use the set of answers determined by the inappropriateness determination unit 62 to satisfy the inappropriateness condition.
- the estimated type obtaining unit 63 obtains the estimated type using the answers to each of M1 questions, which are less than N, out of the answers to the N questions.
- N is a natural number of 2 or more.
- M1 is a natural number of 1 or more.
- the estimation type acquisition unit 63 acquires estimated risk presence/absence information using answers to each of M2 questions, which are less than N, out of the answers to the N questions.
- N is a natural number of 2 or more.
- M2 is a natural number of 1 or more.
- the estimation type acquisition unit 63 acquires an estimation type identifier, for example, using contribution degree acquisition means 631, candidate acquisition means 632, and estimation type acquisition means 633.
- the estimation type acquisition unit 63 may acquire the estimated risk presence/absence information using the contribution degree acquisition means 631 , the candidate acquisition means 632 , and the estimation type acquisition means 633 .
- the contribution acquisition means 631 acquires the contribution of the answers to each of the N questions for the actual type. Note that the degree of contribution of an answer to each question can also be said to be the degree of contribution of each question.
- the contribution degree acquisition means 631 acquires the contribution degree of each question, for example, by the following method (1) or (2). However, the method of acquiring the contribution of each question does not matter.
- the contribution acquisition means 631 acquires, for example, the contribution of the answers to each of the N questions to the actual risk presence/absence information. Note that the degree of contribution of an answer to each question can also be said to be the degree of contribution of each question.
- the contribution degree acquisition means 631 acquires the contribution degree of each question, for example, by the following method (1) or (2). However, the method of acquiring the contribution of each question does not matter. (1) When obtaining the contribution from the random forest learner (1-1) When obtaining the contribution of the question for obtaining the estimation type
- the contribution degree acquisition means 631 uses the random forest module to acquire the contribution degree of the answer.
- the contribution acquisition unit 631 acquires the answer to each question of the question sheet having N questions for each identifier. Further, the contribution degree acquisition unit 631 acquires, for each identifier, the real type paired with the identifier from the real type storage unit 43, for example. Next, the contribution acquisition means 631 constructs a vector whose elements are each answer and the actual type, for example, for each identifier. Next, the contribution degree acquiring means 631, for example, gives vectors corresponding to two or more identifiers to a random forest learning module, executes the random forest learning module, and acquires a learner. Then, the contribution acquisition means 631 acquires the contribution of each question corresponding to each answer using the learning device.
- the degree of contribution of each question is information relating to the contribution of each question to obtaining the actual type, and may be called the contribution rate or the degree of importance. It should be noted that the process of acquiring the degree of contribution of each question corresponding to each answer using a learner acquired by executing the random forest learning module is a known technique. (1-2) When acquiring the contribution of questions for acquiring estimated risk presence/absence information
- the contribution degree acquisition means 631 uses the random forest module to acquire the contribution degree of the answer.
- the contribution acquisition unit 631 acquires the answer to each question of the question sheet having N questions for each identifier. Further, the contribution degree acquisition means 631 acquires real risk presence/absence information paired with the identifier from the real type storage unit 43, for example, for each identifier. Next, the contribution degree acquiring means 631 constructs a vector whose elements are each answer and the actual risk presence/absence information, for example, for each identifier. Next, the contribution degree acquiring means 631, for example, gives vectors corresponding to two or more identifiers to a random forest learning module, executes the random forest learning module, and acquires a learner. Then, the contribution acquisition means 631 acquires the contribution of each question corresponding to each answer using the learning device.
- the degree of contribution of each question is information relating to the contribution of each question for obtaining information on the presence or absence of real risks, and may be called a contribution rate or a degree of importance. It should be noted that the process of acquiring the degree of contribution of each question corresponding to each answer using a learner acquired by executing the random forest learning module is a known technique. (2) When acquiring the contribution using the accuracy evaluation result of the machine learning learner (2-1) When acquiring the contribution of the question for acquiring the estimation type
- the contribution degree acquisition means 631 constructs a vector whose elements are each answer to all questions and the actual type, for example, for each identifier.
- the contribution acquisition means 631 was able to acquire two or more vectors.
- the contribution acquisition means 631 calculates the accuracy (accuracy rate, precision rate, recall rate, F value, etc.) of the learner configured using each answer to all questions and the actual type by k-fold cross validation, for example. acquires the basic accuracy, which is any accuracy among the accuracy of
- the contribution degree acquisition means 631 constructs a vector whose elements are the answers excluding the answer to one question among all the questions and the actual type.
- the contribution degree acquisition means 631 acquires the accuracy of the learning device configured using each answer except the answer to one question and the actual type by, for example, k-fold cross validation, which is the missing accuracy.
- the contribution degree obtaining means 631 calculates the difference between the basic precision and the missing precision.
- the contribution degree acquiring means 631 calculates the degree of contribution of the questions to the excluded answers by, for example, an increasing function with this difference as a parameter.
- the degree of contribution of the question to the excluded answer can be calculated by the arithmetic expression "f (basic precision - missing precision)". Also, the greater the difference between the basic accuracy and the missing accuracy, the greater the contribution of the excluded answer, that is, the excluded question. Moreover, excluding an answer is also excluding a question corresponding to the answer.
- the contribution degree acquisition means 631 changes, for example, the first question to the N-th question to be excluded, and acquires N missing accuracies for each question excluded by k-fold cross-validation.
- the contribution acquisition means 631 performs k-fold cross-validation using a machine learning module.
- the machine learning module may be random forest, decision tree, deep learning, SVR, etc., and any algorithm may be used.
- various machine learning functions such as TensorFlow library, fastText, tinySVM, R language random forest module, and various existing libraries can be used. Since the k-fold cross-validation is a well-known technique, detailed description thereof will be omitted.
- the contribution degree acquisition unit 631 may acquire the contribution degree of each question using only the N missing accuracies, for example, without using the basic accuracy. In other words, the contribution degree acquisition unit 631 may acquire the contribution degree such that the lower the precision indicated by the missing accuracy corresponding to the missing question, the greater the contribution of the question.
- the contribution degree acquiring means 631 may acquire the degree of contribution of each question, for example, by a decreasing function whose parameter is the missing accuracy of each question.
- the contribution degree acquiring means 631 constructs a vector whose elements are each answer to all the questions and the actual risk presence/absence information, for example, for each identifier.
- the contribution acquisition means 631 was able to acquire two or more vectors.
- the contribution degree acquisition means 631 for example, by k-fold cross validation, the accuracy (accuracy rate, precision rate, recall rate, or F Obtain the basic precision, which is any precision of the precision of the value, etc.).
- the contribution degree acquisition means 631 constructs a vector whose elements are, for example, the answers excluding the answer to one question among all the questions and the actual risk presence/absence information, for each identifier.
- the contribution degree acquisition means 631 acquires the missing accuracy, which is the accuracy of the learner configured using each answer except the answer to one question and the actual risk presence/absence information, for example, by k-fold cross-validation.
- the contribution degree obtaining means 631 calculates the difference between the basic precision and the missing precision.
- the contribution degree acquiring means 631 calculates the degree of contribution of the questions to the excluded answers by, for example, an increasing function with this difference as a parameter.
- the degree of contribution of the question to the excluded answer can be calculated by the arithmetic expression "f (basic precision - missing precision)". Also, the greater the difference between the basic accuracy and the missing accuracy, the greater the contribution of the excluded answer, that is, the excluded question. Moreover, excluding an answer is also excluding a question corresponding to the answer.
- the contribution degree acquisition means 631 changes, for example, from the first question to the N-th question to be excluded, and acquires N missing accuracies for each question excluded by k-fold cross-validation.
- the contribution degree acquisition means 631 performs k-fold cross-validation using a machine learning module, as described above.
- the contribution acquisition means 631 may acquire the contribution of each question using only the N missing accuracies, for example, without using the basic accuracies.
- the contribution degree acquisition unit 631 may acquire the contribution degree such that the lower the precision indicated by the missing accuracy corresponding to the missing question, the greater the contribution of the question.
- the contribution degree acquiring means 631 may acquire the degree of contribution of each question, for example, by a decreasing function whose parameter is the missing accuracy of each question.
- the candidate acquisition unit 632 acquires two or more question set candidates.
- a question set candidate is a candidate for a question set.
- a question set is a set of two or more questions with a different number of two or more among the N questions of the question card. It does not matter how the candidate obtaining means 632 obtains two or more question set candidates.
- the candidate acquisition unit 632 acquires two or more question set candidates, for example, by any of the following methods (1) to (3). (1) When using contribution
- the candidate acquisition means 632 acquires, for example, two or more question set candidates, which are sets of questions whose contributions acquired by the contribution acquisition means 631 satisfy a predetermined contribution condition.
- the contribution degree condition is that the contribution degree is large.
- the contribution degree condition is, for example, that the contribution degree is equal to or greater than a threshold value, that the contribution degree is greater than the threshold value, or that the contribution degree is equal to or higher than the top N ranks.
- the candidate acquisition unit 632 acquires, for example, two or more question set candidates that contain the largest number of questions in descending order of contribution.
- Two or more question set candidates in which the number of questions included in descending order of contribution is, for example, "the question with the highest contribution”, "a set of two questions with the first and second contributions", ""A set of 3 questions from 1st to 3rd” ... "A set of X questions from 1st to X”. Note that the number of Xs does not matter.
- the candidate acquisition means 632 for example, from a question sheet that is a set of N questions, N question set candidates for one question, N C 2 question set candidates, N C 3 question set candidates, Question set candidates, N C 4 question set candidates, . That is, the candidate acquiring means 632 obtains ( N C 1 + N C 2 + N C 3 + . . . + N C N-2 + N C N-1 + N C N ) question set candidates are obtained. In such a case, the contribution degree acquiring means 631 is unnecessary.
- the candidate acquisition unit 632 may determine the upper limit of the number of questions included in the set, such as not acquiring a set of N questions. (3) When randomly obtained
- the candidate acquisition means 632 for example, randomly selects a first question set candidate consisting of two questions, and acquires X (X is predetermined, for example). In addition, the candidate acquisition unit 632 randomly selects, for example, second question set candidates consisting of three questions, and acquires X kinds of them. Further, the candidate acquisition unit 632 randomly selects, for example, a third question set candidate consisting of four questions, and acquires X kinds of them. . . . The candidate acquisition unit 632 randomly selects, for example, question set candidates consisting of Y (where Y ⁇ N is predetermined) questions, and acquires X kinds of question set candidates. In such a case, the contribution degree acquiring means 631 is also unnecessary.
- the estimated type acquiring means 633 acquires an estimated type, which is a value obtained using the answers to the two or more questions possessed by each question set candidate.
- the estimated type acquisition unit 633 acquires the estimated type, for example, by the following method (1) or (2).
- the estimated type obtaining means 633 obtains, for each of the two or more question set candidates obtained by the candidate obtaining means 632 and for each identifier, an answer to two or more questions possessed by each question set candidate, and an answer for learning. , from the answer storage unit 42 .
- the estimated type acquisition means 633 acquires from the real type storage unit 43 the actual type corresponding to the answer for learning for each of the two or more question set candidates acquired by the candidate acquisition means 632 and for each user identifier. do.
- the estimated type acquisition means 633 provides the answer and the actual type paired with each of the two or more identifiers to the machine learning learning module. , performs learning processing and acquires a learner. Random forest is suitable for machine learning, but other algorithms such as deep learning, decision tree, and SVR may also be used.
- the estimated type acquisition means 633 obtains the answers to the two or more questions of each question set candidate. An answer is acquired from the answer storage unit 42 . Then, the estimation type acquisition means 633 supplies the acquired learner and test answer for each of the two or more question set candidates and for each identifier to a machine learning prediction module, and executes the prediction module. , to get the estimated type. Note that the estimation type may be called a prediction type or the like.
- the estimated type acquisition means 633 acquires, for each of the two or more question set candidates acquired by the candidate acquisition means 632 and for each identifier, the answers to the two or more questions of each question set candidate from the answer storage unit 42. .
- the estimated type acquisition unit 633 acquires from the actual type storage unit 43 the actual type paired with the identifier for each of the two or more question set candidates acquired by the candidate acquisition unit 632 .
- the estimation type acquisition means 633 acquires an estimation type by k-fold cross-validation for each of the two or more question set candidates and for each identifier. That is, for each of two or more question set candidates, for example, the set of answers of all users and the actual type are divided into k pieces (for example, 10 pieces), and (k ⁇ 1)/k data is processed by the machine. Provide a learning module for learning, execute the learning module, and obtain a learner. Next, the estimation type acquisition means 633 supplies the remaining 1/k data and the learner to a machine learning prediction module, executes the prediction module, and obtains the answer corresponding to the remaining 1/k data. Get the estimated type of the set. Then, the estimation type acquisition means 633 changes data to be given to the learning module and data to be given to the prediction module, executes the above process k times, and obtains an estimation type for each identifier for each of two or more question set candidates. to get
- the estimation type acquisition means 633 acquires estimated risk presence/absence information obtained using answers to two or more questions of each question set candidate for each of the two or more question set candidates acquired by the candidate acquisition means 632.
- the estimation type acquisition unit 633 acquires estimated risk presence/absence information by the following method (1) or (2), for example, in the same way as when acquiring an estimation type.
- (1) A method of configuring a learning device using answers for learning and actual risk presence/absence information, and acquiring estimated risk presence/absence information using the learning device and test answers.
- the estimated type obtaining means 633 obtains, for each of the two or more question set candidates obtained by the candidate obtaining means 632 and for each identifier, an answer to two or more questions possessed by each question set candidate, and an answer for learning. , from the answer storage unit 42 .
- the estimated type acquisition unit 633 stores real risk presence/absence information corresponding to answers for learning for each of the two or more question set candidates acquired by the candidate acquisition unit 632 and for each user identifier. Get from
- the estimation type acquisition means 633 acquires answers paired with the two or more identifiers and the actual risk presence/absence information as a machine learning learning module. , the learning process is performed, and a learner is obtained.
- the estimated type acquisition means 633 obtains the answers to the two or more questions of each question set candidate. An answer is acquired from the answer storage unit 42 . Then, the estimation type acquisition means 633 supplies the acquired learner and test answer for each of the two or more question set candidates and for each identifier to a machine learning prediction module, and executes the prediction module. , to obtain estimated risk presence/absence information.
- the estimated type acquisition means 633 acquires, for each of the two or more question set candidates acquired by the candidate acquisition means 632 and for each identifier, the answers to the two or more questions of each question set candidate from the answer storage unit 42. .
- the estimated type obtaining means 633 obtains real risk presence/absence information paired with the identifier from the actual type storage unit 43 for each of the two or more question set candidates obtained by the candidate obtaining means 632 .
- the estimation type acquisition means 633 acquires estimated risk presence/absence information by k-fold cross-validation for each of the two or more question set candidates and for each identifier. That is, for each of two or more question set candidates, for example, the set of answers of all users and the actual risk presence/absence information are divided into k pieces (for example, 10 pieces), and (k ⁇ 1)/k data is , is given to a learning module of machine learning, the learning module is executed, and a learner is obtained. Next, the estimation type acquisition means 633 supplies the remaining 1/k data and the learner to a machine learning prediction module, executes the prediction module, and obtains the answer corresponding to the remaining 1/k data. Get the estimated risk presence/absence information for the set. Then, the estimation type acquisition means 633 changes the data given to the learning module and the data given to the prediction module, executes the above process k times, and obtains the estimated risk for each identifier for each of two or more question set candidates. Get presence/absence information.
- the question determination unit 64 selects M questions corresponding to the estimation type when the accuracy of the degree of matching between the estimation type acquired by the estimation type acquisition unit 63 and the actual type satisfies a predetermined accuracy condition. Determine questions.
- a set of such M questions is a question sheet.
- the question determination unit 64 determines the M A question that is one question is determined. A set of such M1 questions is the first question sheet.
- the question determination unit 64 selects M 2 questions corresponding to the estimation type. determine a question. A set of such M2 questions is the second question sheet.
- the output unit 7 outputs various information.
- Various types of information are, for example, simple questionnaires, which will be described later.
- output means display on a display, projection using a projector, printing on a printer, sound output, transmission to an external device, storage on a recording medium, and transmission to another processing device or other program.
- the simple question sheet output unit 71 outputs a simple question sheet that is information about the question set, which is a set of M questions determined by the question determination unit 64.
- the simple question sheet may be a set of M questions, a set of M question IDs, a set of link information to each M questions, or the like.
- the simple question sheet may be any information that allows acquisition of a set of M questions.
- the simplified question sheet to be output may be a question obtained by processing at least one of the M questions determined by the question determination unit 64 .
- the simple question sheet may be the first question sheet or the second question sheet described above.
- the storage unit 4, the question sheet storage unit 41, the answer storage unit 42, and the actual type storage unit 43 are preferably non-volatile recording media, but can also be realized with volatile recording media.
- information may be stored in the storage unit 4 or the like via a recording medium, or information transmitted via a communication line or the like may be stored in the storage unit 4 or the like.
- information input via an input device may be stored in the storage unit 4 or the like.
- the reception unit 5 and the answer reception unit 51 can be realized by device drivers for input means such as touch panels and keyboards, control software for menu screens, and the like. Note that the reception unit 5 and the like may be realized by wireless or wired communication means.
- the processing unit 6, the actual type acquiring unit 61, the inappropriateness determining unit 62, the presumed type acquiring unit 63, the question determining unit 64, the contribution degree acquiring unit 631, the candidate acquiring unit 632, and the presumed type acquiring unit 633 usually include a processor, It can be implemented from memory or the like.
- the processing procedure of the processing unit 6 and the like is normally realized by software, and the software is recorded in a recording medium such as a ROM. However, it may be realized by hardware (dedicated circuit).
- the processor is, for example, a CPU, MPU, GPU, or the like, and any type of processor does not matter.
- the output unit 7 and simple question form output unit 71 may or may not include output devices such as displays and speakers.
- the output unit 7 and the like can be realized by output device driver software, or by output device driver software and an output device. Note that the output unit 7 and the like may be realized by wireless or wired communication means. Also, the output unit 7 and the like may be implemented by a processor, memory, or the like.
- Step S1401 The response receiving unit 51 determines whether or not the response to each of the N questions that make up the question sheet and the measured value have been received in pairs with the identifier.
- the measured value is the actual type or risk presence/absence information. If an answer or the like has been accepted, the process goes to step S1402, and if no answer or the like has been accepted, the process goes to step S1403.
- the identifier is, for example, a user identifier. Also, the identifiers are, for example, a user identifier and a day identifier.
- Step S1402 The processing unit 6 stores the set of answers accepted in step S201 and the measured values in the answer storage unit 42 in pairs with identifiers. Return to step S1401.
- Step S1403 The reception unit 5 determines whether or not a simple question sheet output instruction has been received. If the simple question sheet output instruction has been received, the process proceeds to step S1404, and if the simple question sheet output instruction has not been received, the process returns to step S1401.
- Step S1404 The processing unit 6 performs question determination processing.
- An example of question determination processing will be described with reference to the flowcharts of FIGS. 15 and 18.
- FIG. The question determination process is a process of determining questions for constructing the simple question sheet.
- Step S1405 The processing unit 6 uses the one or more questions determined in step S1404 to acquire a simplified question sheet to be output.
- Step S1406 The simple question sheet output unit 71 outputs the simple question sheet acquired in step S1407. Return to step S1401. Here, the simple question sheet output unit 71 accumulates the simple question sheet acquired in step S1407 in the question sheet storage unit 41, for example.
- step S1404 Next, a first example of question determination processing in step S1404 will be described using the flowchart of FIG.
- Step S1501 The contribution acquisition means 631 acquires the contribution of each question in the question sheet. An example of such contribution degree acquisition processing will be described with reference to the flowcharts of FIGS. 16 and 17. FIG.
- Step S1502 The estimation type acquisition unit 63 substitutes 1 for the counter i.
- Step S1503 The candidate acquisition unit 632 acquires the i-th question set candidate using the contribution of each question acquired in step S1501.
- the candidate acquisition unit 632 acquires i questions from the question sheet of the question sheet storage unit 41, from the question with the highest degree of contribution to the question with the i-th largest degree of contribution, for example.
- Step S1504 The estimation type acquisition means 633 acquires learning data from the answer storage unit 42.
- the estimated type acquisition means 633 acquires from the answer storage unit 42 a set of answers to each question of the i-th question set candidate, which is a set of answers paired with the usage "for learning”. In addition, the estimated type acquisition unit 633 acquires from the actual type storage unit 43 the actual measurement value paired with each identifier for each identifier paired with the usage “for learning”. Next, the estimated type acquisition means 633 acquires a vector having, as elements, each answer to each question and an actual measurement value for each identifier. Such learning data is a set of vectors corresponding to each of two or more identifiers.
- Step S1505 The estimation type acquisition means 633 supplies the learning data acquired in step S1504 to the learning processing module of machine learning, executes the module, and acquires a learner.
- the machine learning module is, for example, a random forest module.
- Step S1506 The estimation type acquisition means 633 substitutes 1 for the counter j.
- Step S1507 The estimation type acquisition means 633 determines whether the j-th test data is stored in the answer storage unit 42 or not. If the j-th test data exists, the process goes to step S1508, and if the j-th test data does not exist, the process goes to step S1511.
- the test data is a set of answers associated with identifiers.
- Step S1508 The estimated type acquisition means 633 selects each answer of each question of the i-th question set candidate, which is the j-th test data, from among the test data stored in the answer storage unit 42. A set of answers having elements is acquired from the answer storage unit 42 . Next, the estimation type acquisition unit 633 provides the set of answers and the learner acquired in step S1505 to a machine learning prediction module, executes the prediction module, and acquires an estimated value.
- the estimated value is the estimated type or the estimated risk presence/absence information.
- Step S1509 The question determination unit 64 acquires from the answer storage unit 42 an identifier paired with the j-th test data. Next, the question determination unit 64 acquires the measured value paired with the identifier from the actual type storage unit 43, and associates the measured value with the estimated value acquired in step S410.
- Step S1510 The estimated type acquisition means 633 increments the counter j by 1. Return to step S1507.
- Step S1511 The question determining unit 64 acquires sets of estimated values acquired in step S1508 and measured values acquired in step S409 for the number of test data. Next, the question determining section 64 acquires accuracy information regarding the accuracy of the estimated value with respect to the actual measured value, using all the information of the set of the estimated value and the actual measured value.
- the question determination unit 64 uses, for example, all the information of pairs of estimated values and measured values to acquire accuracy information, which is the proportion of matches between estimated values and measured values.
- the question determination unit 64 uses, for example, all the information of pairs of estimated values and measured values to obtain accuracy information, which is the correlation coefficient between the estimated values and the measured values.
- Step S1512 The question determination unit 64 determines whether the accuracy information acquired in step S413 satisfies the approximation condition stored in the storage unit 4. If the approximation condition is satisfied, the process goes to step S1513, and if the approximation condition is not satisfied, the process goes to step S1514.
- the approximation condition is, for example, that the accuracy information is greater than or equal to a threshold value, or that the accuracy information is greater than a threshold value.
- the question determination unit 64 acquires information about the i-th question set candidate. Return to upper process.
- the information about the i-th question set candidate is, for example, the i-th question set candidate itself or the question identifier of each question of the i-th question set candidate.
- Step S1514 The estimated type acquisition unit 63 increments the counter i by 1. Return to step S1503.
- step S1501 a first example of contribution degree acquisition processing in step S1501 will be described using the flowchart of FIG.
- Step S1601 The contribution acquisition means 631 substitutes 1 for the counter i.
- Step S1602 The contribution acquisition means 631 determines whether or not the i-th identifier exists. If the i-th identifier exists, go to step S1603, and if the i-th identifier does not exist, go to step S1607.
- Step S1603 The contribution acquisition means 631 acquires answers to all questions corresponding to the i-th identifier.
- Step S1604 The contribution degree acquisition means 631 acquires the measured value corresponding to the i-th identifier from the actual type storage unit 43. Note that the actual measurement value is the actual type or the actual risk presence/absence information.
- Step S1605 The contribution degree acquisition means 631 constructs a vector whose elements are all the answers acquired in step S1603 and the measured values acquired in step S1604.
- Step S1606 The contribution acquisition means 631 increments the counter i by 1. Return to step S1602.
- Step S1607 The contribution acquisition means 631 gives all the vectors configured in step S1605 to the random forest learning module, executes the learning module, and acquires a learner.
- the learning device is a learning device that can be used for prediction processing of a random forest, and is used to input the answers to all the questions corresponding to the type identifier of the type of interest and to output the estimated value. It is a learner that
- Step S1608 The contribution acquisition means 631 substitutes 1 for the counter j.
- Step S1609 The contribution acquisition means 631 determines whether or not the j-th question exists. If the j-th question exists, go to step S1610, and if the j-th question does not exist, return to the upper process.
- Step S1610 The contribution acquisition means 631 acquires the contribution of the j-th question (j-th feature amount) using the learner acquired in step S1607. Note that the process of acquiring the contribution of the j-th feature amount using a random forest learner is a known technique.
- Step S1611 The contribution degree acquisition means 631 accumulates the contribution degree acquired in step S1610 in association with the j-th question.
- Step S1612 The contribution acquisition means 631 increments the counter j by 1. Return to step S1609.
- step S1501 an example of the second contribution degree acquisition process in step S1501 will be described using the flowchart of FIG. In the flowchart of FIG. 17, description of the same steps as in the flowchart of FIG. 16 will be omitted.
- Step S1701 Contribution acquisition means 631 uses the set of vectors acquired in step S1605 to perform k-fold cross-validation using a machine learning algorithm. to get a base precision of .
- Step S1702 The contribution degree acquisition means 631 temporarily accumulates the basic accuracy acquired in step S1701.
- Step S1703 The contribution acquisition means 631 substitutes 1 for the counter j.
- Step S1704 The contribution degree acquisition means 631 determines whether or not the j-th question exists among all the target questions. If the j-th question exists, go to step S1705, and if the j-th question does not exist, return to the higher-level processing.
- Step S1705 The contribution acquisition means 631 substitutes 1 for the counter k.
- Step S1706 The contribution acquisition means 631 determines whether or not the k-th identifier exists. If the k-th identifier exists, go to step S1707, and if the k-th identifier does not exist, go to step S1709.
- the contribution degree acquisition means 631 acquires the vector that is the vector acquired in step S1605 and pairs with the k-th identifier, and constructs a vector that excludes the answer to the j-th question from this vector. In other words, the contribution degree acquisition unit 631 constructs a vector whose dimension is one less than the vector acquired in step S1605.
- Step S1708 The contribution acquisition means 631 increments the counter k by 1. Return to step S1706.
- Step S1709 Contribution degree acquisition means 631 performs k-fold cross-validation using a machine learning algorithm using the set of vectors acquired in step S1707, and uses the answer excluding the answer to the j-th question. Get the missing precision, which is the precision of .
- Step S1710 The contribution degree acquisition means 631 temporarily accumulates the missing accuracy acquired in step S1709 in association with the j-th question.
- Step S1711 The contribution degree acquisition means 631 acquires the difference between the missing accuracy temporarily accumulated in step S1702 and the basic accuracy temporarily accumulated in step S1710. The greater the difference, the greater the degree of contribution of the answer to the j-th question.
- Step S1712 The contribution degree acquisition means 631 accumulates the contribution degree acquired in step S1711 in association with the j-th question.
- Step S1713 The contribution acquisition means 631 increments the counter j by 1. Return to step S1704.
- step S1604 a second example of question determination processing in step S1604 will be described using the flowchart of FIG.
- Step S1801 The question determination unit 64 substitutes 1 for the counter j.
- Step S1802 The question determination unit 64 substitutes 1 for the counter k.
- Step S1803 The question determining unit 64 determines whether or not there is a k-th combination of j questions. If the k-th combination exists, go to step S1804; if not, go to step S1814.
- the question determination unit 64 acquires a question set candidate that is the k-th combination of j questions.
- Step S1805) The question determination unit 64 substitutes 1 for the counter l.
- Step S1806 The question determination unit 64 determines whether or not the l-th identifier exists. If the l-th identifier exists, go to step S1807, and if the l-th identifier does not exist, go to step S1811.
- Step S1807 The question determination unit 64 acquires from the answer storage unit 42 the answer to each question of the question set candidate acquired in step S1804, which is paired with the l-th identifier.
- Step S1808 The question determination unit 64 acquires from the actual type storage unit 43 the actual value corresponding to the i-th type and paired with the l-th identifier.
- Step S1809 The question determination unit 64 constructs a vector whose elements are the answers to the questions acquired in step S1807 and the actual values acquired in step S1808.
- Step S1810 The counter l is incremented by 1. Return to step S1806.
- Step S1811 The question determination unit 64 performs k-fold cross-validation using the set of vectors acquired in step S1809, and determines the precision of the learner configured using the set of vectors acquired in step S1809. Get information.
- Step S1812 The question determination unit 64 stores the accuracy information acquired in step S1811 in a buffer (not shown) in association with the k-th question set candidate.
- Step S1813 The question determination unit 64 increments the counter k by 1. Return to step S1803.
- Step S1814 The question determination unit 64 acquires the best accuracy information when j questions are used from among the accuracy information accumulated in step S1812.
- Step S1815 The question determination unit 64 determines whether or not the accuracy information acquired in step S1814 satisfies the approximation condition of the storage unit 4. If the approximation condition is satisfied, the process goes to step S1816, and if the approximation condition is not satisfied, the process goes to step S1817.
- Step S1816 The question determination unit 64 acquires information about the k-th question set candidate of the j questions. Go to step S416. Note that the k-th question set candidate of the j questions is the set of questions determined by the question determination unit 64, and is the set of questions used to compose the simple question sheet.
- Step S1817 The question determination unit 64 increments the counter j by 1. Return to step S1802.
- the question sheet shown in FIG. 19 is now stored in the question sheet storage unit 41.
- the question sheet here has 350 question information.
- the answer storage unit 42 stores the answers to the question sheets of 434 users. Such a response is shown in FIG. In FIG. 20, "ID" is the identifier of the user who responded. In addition, in FIG. 20, each answer is associated with "use”. Here, the usage is one of "for learning”, “for verification”, and “for testing”.
- the “for learning” record is used for constructing random forest learners.
- the “validation” record is used to adjust various parameters of the random forest learner.
- “Test” records are used to obtain estimated values, obtain accuracy information with actual values, and determine final questions. In addition, the "usage” may be added automatically, or may be added manually.
- the actual type storage unit 43 stores the actual risk presence/absence information and the actual type in association with the user's identifier.
- the contribution acquisition means 631 of the simplified question sheet acquisition device B acquires the contribution of each question when acquiring the estimated risk presence/absence information by the above-described algorithm of FIG. 16 or FIG.
- the degree of contribution is temporarily stored in association with each question.
- the candidate acquisition means 632 and the question determination unit 64 determine questions for the simple question sheet when acquiring estimated risk presence/absence information, using the algorithm in FIG.
- the determined questions are, for example, the 15 questions in FIG.
- the question determining unit 64 and the like may determine the questions of the simple question sheet when acquiring the estimated risk presence/absence information by the algorithm of FIG.
- the contribution degree acquiring means 631 of the simple question sheet acquisition device B acquires the contribution degree of each question when acquiring the estimated type by the above-described algorithm of FIG. 16 or FIG. Temporarily accumulate contribution.
- the candidate acquisition means 632 and the question determination unit 64 determine the questions of the simple question sheet when acquiring the estimated type using the algorithm in FIG.
- the determined questions are, for example, the eight questions in FIG.
- the question determination unit 64 and the like may determine the questions of the simple question sheet when acquiring the estimation type by the algorithm of FIG.
- the processing in this embodiment may be realized by software. Then, this software may be distributed by software download or the like. Also, this software may be recorded on a recording medium such as a CD-ROM and distributed. Note that this also applies to other embodiments in this specification.
- the software that implements the simple question sheet acquisition device B in this embodiment is the following program. In other words, this program provides the computer with answers to N questions, M answers corresponding to either two or three or more real type identifiers, which are less than the N questions. Presumed type acquisition unit for acquiring presumed type identifiers obtained by using answers to each question, and accuracy in which the degree of matching between the presumed type identifiers acquired by the presumed type acquisition unit and the actual type identifiers is determined in advance. a question determination unit that determines M questions corresponding to the presumed type identifier when a condition is satisfied; This is a program for functioning as a simple question sheet output unit that outputs a question sheet.
- FIG. 21 shows the appearance of a computer that executes the program described in this specification and realizes the learning device 1, the blood glucose constitution determination device 2, the simple questionnaire acquisition device B, etc. of the various embodiments described above. show.
- the embodiments described above may be implemented in computer hardware and computer programs running thereon.
- FIG. 21 is an overview diagram of this computer system 300
- FIG. 22 is a block diagram of the system 300. As shown in FIG.
- computer system 300 includes computer 301 including a CD-ROM drive, keyboard 302 , mouse 303 and monitor 304 .
- a computer 301 includes a CD-ROM drive 3012, an MPU 3013, a bus 3014 connected to the CD-ROM drive 3012, a ROM 3015 for storing programs such as a boot-up program, It includes a RAM 3016 connected and for temporarily storing application program instructions and providing temporary storage space, and a hard disk 3017 for storing application programs, system programs and data.
- computer 301 may also include a network card that provides connection to a LAN.
- a program that causes the computer system 300 to execute the functions of the blood sugar constitution determination device 2 of the above-described embodiment is stored in the CD-ROM 3101, inserted into the CD-ROM drive 3012, and transferred to the hard disk 3017. good.
- the program may be transmitted to computer 301 via a network (not shown) and stored in hard disk 3017 .
- Programs are loaded into RAM 3016 during execution.
- the program may be loaded directly from CD-ROM 3101 or network.
- the program does not necessarily include an operating system (OS) that causes the computer 301 to execute the functions of the blood glucose constitution determination device 2 of the above-described embodiment, or a third party program.
- OS operating system
- a program need only contain those parts of instructions that call the appropriate functions (modules) in a controlled manner to produce the desired result. How the computer system 300 operates is well known and will not be described in detail.
- the step of transmitting information, the step of receiving information, etc. are performed by hardware. processing) are not included.
- the computer that executes the above program may be singular or plural. That is, centralized processing may be performed, or distributed processing may be performed.
- two or more communication means existing in one device may be physically realized in one medium.
- each process may be implemented by centralized processing by a single device, or may be implemented by distributed processing by a plurality of devices.
- the blood sugar constitution determination device has the effect that the user's blood sugar constitution type can be easily known, and is useful as a blood sugar constitution determination device or the like.
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Public Health (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Epidemiology (AREA)
- Primary Health Care (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Physics & Mathematics (AREA)
- Pathology (AREA)
- Databases & Information Systems (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Optics & Photonics (AREA)
- Emergency Medicine (AREA)
- Biophysics (AREA)
- Heart & Thoracic Surgery (AREA)
- Molecular Biology (AREA)
- Surgery (AREA)
- Animal Behavior & Ethology (AREA)
- Veterinary Medicine (AREA)
- Computational Linguistics (AREA)
- Medical Treatment And Welfare Office Work (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
Description
本実施の形態において、生活習慣に関する設問を含むアンケートに含まれる1以上の各設問に対する回答を用いて、血糖体質のタイプを取得し、出力する血糖体質判定装置について説明する。なお、血糖体質のタイプとは、血糖に関する健康状態に関するタイプである。血糖体質のタイプは、例えば、インスリン感受性に関するタイプ(抵抗性なし,抵抗性傾向ありのいずれか)、インスリン分泌に関するタイプ(正常、出にくいのいずれか)の組み合わせの4タイプのうちの一のタイプである。血糖体質のタイプは、例えば、血糖値が上がりやすいタイプか否か、血糖値が下がりやすいタイプか否かである。血糖体質のタイプの数、種類等は問わない。アンケートは、直接的に血糖値に影響を及ばさないと考えられる事項に関する設問を含むことが好適である。アンケートは、回答し易い設問が多いことが好適である。また、生活習慣に関する設問は、例えば、食生活の習慣に関する設問、飲水または飲酒習慣に関する設問、運動習慣または生活強度に関する設問、睡眠または眠気に関する設問、衛生に関する設問、その他生活習慣に関する一般的な設問である。一般的な設問は、例えば、歯磨きの頻度に関する設問、階段の利用度に関する設問である。
(1)元情報が学習器である場合
(1-1)元情報が、一度の予測処理により血糖体質のタイプを決定する学習器である場合
(1-2)元情報が、血糖に関するリスクの有無を判断するための第一学習器と、リスクが有る場合に血糖体質のタイプを決定するための第二学習器とである場合
(2)元情報が対応表である場合
(2-1)元情報が、血糖体質のタイプを決定する一の対応表である場合
(2-2)元情報が、血糖に関するリスクの有無を判断するための第一対応表と、リスクが有る場合に血糖体質のタイプを決定するための第二対応表とである場合
本実施の形態において、実施の形態1で述べた設問票、第一設問票、または第二設問票を取得する簡易設問票取得装置Bについて説明する。
(1)ランダムフォレストの学習器から寄与度を取得する場合
(1-1)推定タイプを取得するための設問の寄与度を取得する場合
(1-2)推定リスク有無情報を取得するための設問の寄与度を取得する場合
(2)機械学習の学習器の精度の評価結果を用いて寄与度を取得する場合
(2-1)推定タイプを取得するための設問の寄与度を取得する場合
(2-2)推定リスク有無情報を取得するための設問の寄与度を取得する場合
(1)寄与度を用いる場合
(2)全組み合わせを取得する場合
(3)ランダムに取得する場合
(1)学習のための回答および実タイプを用いて学習器を構成し、当該学習器とテスト用の回答を用いて推定タイプ識別子を取得する方法
(2)k分割交差検証による方法
(1)学習のための回答および実リスク有無情報を用いて学習器を構成し、当該学習器とテスト用の回答を用いて推定リスク有無情報を取得する方法
(2)k分割交差検証による方法
Claims (13)
- 血糖体質のタイプを判断するための1以上の設問情報であり、生活習慣に関する設問を含む1以上の設問情報が格納される格納部と、
前記1以上の設問情報を出力する出力部と、
前記1以上の各設問情報に対する回答情報をユーザから受け付ける受付部と、
前記受付部が受け付けた前記1以上の回答情報を用いて、前記ユーザの血糖体質のタイプを決定する処理部と、
前記血糖体質のタイプに関する出力情報を出力する情報出力部とを具備する血糖体質判定装置。 - 前記格納部は、
血糖に関するリスクの有無を判断するための1以上の第一設問であり、生活習慣に関する設問を含む1以上の第一設問が格納される第一設問格納部と、
血糖体質のタイプを判断するための1以上の第二設問であり、生活習慣に関する設問を含む1以上の第二設問が格納される第二設問格納部とを具備し、
前記設問情報は、前記第一設問または前記第二設問であり、
前記受付部は、
前記1以上の各第一設問に対する第一回答をユーザから受け付ける第一回答受付部と、
前記1以上の各第二設問に対する第二回答を前記ユーザから受け付ける第二回答受付部とを具備し、
前記処理部は、
前記第一回答受付部が受け付けた前記1以上の第一回答を用いて、前記ユーザの血糖に関するリスクの有無を判断する第一判断部と、
前記第二回答受付部が受け付けた前記1以上の第二回答を用いて、前記ユーザの血糖体質のタイプを決定する第二判断部とを具備し、
前記出力部は、
前記1以上の第一設問を出力する第一設問出力部と、
前記第一判断部がリスク有りと判断した場合に、前記1以上の第二設問を出力する第二設問出力部とを具備する請求項1記載の血糖体質判定装置。 - 前記第二判断部は、
前記第一回答受付部が受け付けた前記1以上の第一回答をも用いて、前記ユーザの血糖体質のタイプを決定する、請求項2記載の血糖体質判定装置。 - 1以上の回答情報と血糖体質のタイプとを有する2以上の教師データに対して、機械学習の学習処理により取得された学習器が格納される学習器格納部をさらに具備し、
前記処理部は、
前記受付部が受け付けた前記1以上の回答情報と前記学習器とを用いて、機械学習の予測処理により血糖体質のタイプを取得する、請求項1記載の血糖体質判定装置。 - 1以上の第二回答と血糖体質のタイプとを有する2以上の教師データに対して、機械学習の学習処理により取得された学習器が格納される学習器格納部をさらに具備し、
前記第二判断部は、
前記第二回答受付部が受け付けた前記1以上の第二回答と前記学習器とを用いて、機械学習の予測処理により血糖体質のタイプを取得する、請求項2記載の血糖体質判定装置。 - 前記血糖体質のタイプを識別する1以上の各タイプ識別子に対応付けて、1以上のアドバイス情報が格納されるアドバイス格納部をさらに具備し、
前記情報出力部は、
前記第二判断部が決定した前記タイプを識別するタイプ識別子に対応付いている1以上のアドバイス情報を含む出力情報を出力する、請求項2記載の血糖体質判定装置。 - 前記第二判断部が決定した前記タイプを識別するタイプ識別子に対応付いている1以上のアドバイス情報を前記アドバイス格納部から取得し、前記タイプ識別子と前記1以上のアドバイス情報とを有する出力情報を構成する出力情報構成部をさらに具備し、
前記情報出力部は、
前記出力情報構成部が構成した出力情報を出力する、請求項6記載の血糖体質判定装置。 - 前記格納部の前記1以上の設問情報は、簡易設問票取得装置が取得した設問であり、
前記簡易設問票取得装置は、
N個の設問に対する回答であり、2または3以上の実タイプ識別子のうちのいずれかの実タイプ識別子に対応する回答のうち、前記N個より少ないM個の各設問に対する回答を用いて得られる推定タイプ識別子を取得する推定タイプ取得部と、
前記推定タイプ取得部が取得した推定タイプ識別子と前記実タイプ識別子との一致の度合いに関する精度が予め決められた精度条件を満たす場合に、当該推定タイプ識別子に対応するM個の設問である設問を決定する設問決定部と、
前記設問決定部が決定したM個の設問の集合である設問集合に関する情報である簡易設問票を出力する簡易設問票出力部とを具備する、請求項1記載の血糖体質判定装置。 - 前記第一設問格納部の前記1以上の第一設問は、簡易設問票取得装置が取得した設問であり、
前記簡易設問票取得装置は、
N個の設問に対する回答であり、前記リスクが有るか無いかのいずれかを特定する実リスク有無情報に対応する前記N個の回答のうち、前記N個より少ないM個の各設問に対する回答を用いて得られる推定リスク有無情報を取得する推定タイプ取得部と、
前記推定タイプ取得部が取得した推定リスク有無情報と前記実リスク有無情報との一致の度合いに関する精度が予め決められた精度条件を満たす場合に、当該推定リスク有無情報に対応するM個の設問である設問を決定する設問決定部と、
前記設問決定部が決定したM個の設問の集合である設問集合に関する情報である簡易設問票を出力する簡易設問票出力部とを具備する、請求項2または請求項3記載の血糖体質判定装置。 - 前記第二設問格納部の前記1以上の第二設問のうちの1以上の種類の設問は、簡易設問票取得装置が取得した設問であり、
前記簡易設問票取得装置は、
N個の設問に対する回答であり、2または3以上の実タイプ識別子のうちのいずれかの実タイプ識別子に対応する回答のうち、前記N個より少ないM個の各設問に対する回答を用いて得られる推定タイプ識別子を取得する推定タイプ取得部と、
前記推定タイプ取得部が取得した推定タイプ識別子と前記実タイプ識別子との一致の度合いに関する精度が予め決められた精度条件を満たす場合に、当該推定タイプ識別子に対応するM個の設問である設問を決定する設問決定部と、
前記設問決定部が決定したM個の設問の集合である設問集合に関する情報である簡易設問票を出力する簡易設問票出力部とを具備する、請求項2または請求項3記載の血糖体質判定装置。 - 前記血糖体質のタイプは、インスリン感受性に関するタイプとインスリン分泌に関するタイプとの組み合わせたタイプである、請求項1から請求項10いずれか一項に記載の血糖体質判定装置。
- 血糖体質のタイプを判断するための1以上の設問情報であり、生活習慣に関する設問を含む1以上の設問情報が格納される格納部と、出力部と、受付部と、処理部と、情報出力部とにより実現される血糖体質判定方法であって、
前記出力部が、前記1以上の設問情報を出力する出力ステップと、
前記受付部が、前記1以上の各設問情報に対する回答情報をユーザから受け付ける受付ステップと、
前記処理部が、前記受付部が受け付けた前記1以上の回答情報を用いて、前記ユーザの血糖体質のタイプを決定する処理ステップと、
前記情報出力部が、前記血糖体質のタイプに関する出力情報を出力する情報出力ステップとを具備する血糖体質判定方法。 - 血糖体質のタイプを判断するための1以上の設問情報であり、生活習慣に関する設問を含む1以上の設問情報が格納される格納部にアクセス可能なコンピュータを、
前記1以上の設問情報を出力する出力部と、
前記1以上の各設問情報に対する回答情報をユーザから受け付ける受付部と、
前記受付部が受け付けた前記1以上の回答情報を用いて、前記ユーザの血糖体質のタイプを決定する処理部と、
前記血糖体質のタイプに関する出力情報を出力する情報出力部として機能させるためのプログラムを記録した記録媒体。
Priority Applications (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP21937955.9A EP4328922A1 (en) | 2021-04-20 | 2021-08-17 | Blood sugar constitution assessment device, blood sugar constitution assessment method, and recording medium |
CN202180097107.0A CN117280421A (zh) | 2021-04-20 | 2021-08-17 | 血糖体质判定装置、血糖体质判定方法及存储介质 |
US18/287,123 US20240127911A1 (en) | 2021-04-20 | 2021-08-17 | Blood sugar constitution determination device, blood sugar constitution determination method, and recording medium |
KR1020237039616A KR20230173690A (ko) | 2021-04-20 | 2021-08-17 | 혈당 체질 판정 장치, 혈당 체질 판정 방법 및 기록 매체 |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2021-071175 | 2021-04-20 | ||
JP2021071175A JP7009663B1 (ja) | 2021-04-20 | 2021-04-20 | 血糖体質判定装置、血糖体質判定方法、およびプログラム |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2022224465A1 true WO2022224465A1 (ja) | 2022-10-27 |
Family
ID=80629677
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2021/030047 WO2022224465A1 (ja) | 2021-04-20 | 2021-08-17 | 血糖体質判定装置、血糖体質判定方法、および記録媒体 |
Country Status (7)
Country | Link |
---|---|
US (1) | US20240127911A1 (ja) |
EP (1) | EP4328922A1 (ja) |
JP (1) | JP7009663B1 (ja) |
KR (1) | KR20230173690A (ja) |
CN (1) | CN117280421A (ja) |
TW (1) | TW202242906A (ja) |
WO (1) | WO2022224465A1 (ja) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2024122536A1 (ja) * | 2022-12-07 | 2024-06-13 | サントリーホールディングス株式会社 | 食後高血糖タイプ判定装置、食後高血糖タイプ判定方法、およびプログラム |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2016048531A (ja) | 2014-08-28 | 2016-04-07 | テルモ株式会社 | 血糖値情報管理装置、血糖値情報管理システム及び血糖値改善促進シート |
CN106923800A (zh) * | 2015-12-31 | 2017-07-07 | 天津森宇科技发展有限公司 | 基于人工智能的智慧健康装置 |
CN111159528A (zh) * | 2019-03-19 | 2020-05-15 | 深圳碳云智能数字生命健康管理有限公司 | 问卷信息的推送方法及装置、存储介质和电子装置 |
-
2021
- 2021-04-20 JP JP2021071175A patent/JP7009663B1/ja active Active
- 2021-08-17 CN CN202180097107.0A patent/CN117280421A/zh active Pending
- 2021-08-17 EP EP21937955.9A patent/EP4328922A1/en active Pending
- 2021-08-17 US US18/287,123 patent/US20240127911A1/en active Pending
- 2021-08-17 WO PCT/JP2021/030047 patent/WO2022224465A1/ja active Application Filing
- 2021-08-17 KR KR1020237039616A patent/KR20230173690A/ko active Search and Examination
- 2021-08-18 TW TW110130516A patent/TW202242906A/zh unknown
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2016048531A (ja) | 2014-08-28 | 2016-04-07 | テルモ株式会社 | 血糖値情報管理装置、血糖値情報管理システム及び血糖値改善促進シート |
CN106923800A (zh) * | 2015-12-31 | 2017-07-07 | 天津森宇科技发展有限公司 | 基于人工智能的智慧健康装置 |
CN111159528A (zh) * | 2019-03-19 | 2020-05-15 | 深圳碳云智能数字生命健康管理有限公司 | 问卷信息的推送方法及装置、存储介质和电子装置 |
Non-Patent Citations (1)
Title |
---|
TAN, SAKU ET AL.: "Development of a questionnaire for insulin resistance diagnosis", THE JOURNAL OF ADULT DISEASES, vol. 43, no. 5, 1 January 2013 (2013-01-01), pages 632, XP009540695, ISSN: 1347-0418 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2024122536A1 (ja) * | 2022-12-07 | 2024-06-13 | サントリーホールディングス株式会社 | 食後高血糖タイプ判定装置、食後高血糖タイプ判定方法、およびプログラム |
Also Published As
Publication number | Publication date |
---|---|
KR20230173690A (ko) | 2023-12-27 |
CN117280421A (zh) | 2023-12-22 |
TW202242906A (zh) | 2022-11-01 |
US20240127911A1 (en) | 2024-04-18 |
JP2022165712A (ja) | 2022-11-01 |
JP7009663B1 (ja) | 2022-01-25 |
EP4328922A1 (en) | 2024-02-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Schrempft et al. | The obesogenic quality of the home environment: associations with diet, physical activity, TV viewing, and BMI in preschool children | |
JP5870328B1 (ja) | 健康管理サーバおよび健康管理サーバ制御方法並びに健康管理プログラム | |
Waki et al. | DialBetics with a multimedia food recording tool, FoodLog: smartphone-based self-management for type 2 diabetes | |
Wang et al. | The role of patient perception of crowding in the determination of real-time patient satisfaction at Emergency Department | |
Monteiro et al. | Secure base representations for both fathers and mothers predict children's secure base behavior in a sample of Portuguese families | |
CN103150464A (zh) | 认知症信息输出装置、认知症看护支援方法和系统 | |
US7634538B2 (en) | Product counseling system, product development program, and machine-readable recording medium | |
US20170242971A1 (en) | Characteristic evaluation apparatus, characteristic evaluation system, characteristic evaluation method, and characteristic evaluation program | |
WO2022224465A1 (ja) | 血糖体質判定装置、血糖体質判定方法、および記録媒体 | |
Fermont et al. | Risk assessment for hospital admission in patients with COPD; a multi-centre UK prospective observational study | |
Niva | Online weight-loss services and a calculative practice of slimming | |
WO2003071937A1 (fr) | Systeme de soin esthetique | |
Brewster et al. | Feasibility and psychometric integrity of mobile phone-based intensive measurement of cognition in older adults | |
EP3886107A1 (en) | Document creation device, method and program | |
EP4318320A1 (en) | Model generation method, computer program, information processing device, information processing system, information processing method, and method for generating training data | |
KR102214792B1 (ko) | 식생활습관과 건강상태 자가진단 체크 시스템 및 이를 이용한 맞춤형 건강정보 제공방법 | |
JP7153893B2 (ja) | 健康指導支援装置および健康指導支援方法 | |
JP6804489B2 (ja) | 情報提供装置及び情報提供方法 | |
JP7193011B2 (ja) | 摂取栄養量推定システム、摂取栄養量推定方法、摂取栄養量推定装置及び摂取栄養量出力方法 | |
JP7515389B2 (ja) | 簡易質問票取得装置、簡易質問票の生産方法、およびプログラム | |
JP2022092521A (ja) | 水分アドバイス出力装置、水分アドバイス出力方法、およびプログラム | |
JP7352761B2 (ja) | 専門家回答予測システム | |
WO2023162829A1 (ja) | 情報処理装置、情報処理方法、および記録媒体 | |
JP2022092520A (ja) | 水分摂取量推定装置、水分摂取量推定方法、およびプログラム | |
WO2024122536A1 (ja) | 食後高血糖タイプ判定装置、食後高血糖タイプ判定方法、およびプログラム |
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: 21937955 Country of ref document: EP Kind code of ref document: A1 |
|
WWE | Wipo information: entry into national phase |
Ref document number: 202180097107.0 Country of ref document: CN Ref document number: 18287123 Country of ref document: US |
|
ENP | Entry into the national phase |
Ref document number: 20237039616 Country of ref document: KR Kind code of ref document: A |
|
WWE | Wipo information: entry into national phase |
Ref document number: 1020237039616 Country of ref document: KR |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2021937955 Country of ref document: EP |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
ENP | Entry into the national phase |
Ref document number: 2021937955 Country of ref document: EP Effective date: 20231120 |