Diagnostic assistance systems and diagnostic assistance devices

Abstract

The diagnosis support system (100) includes a determination unit (21) which receives learning model information (M) generated by a learning model generation unit (11) via an external network (30), and determines whether a patient (P2) not included in a patient group (PF) has a disease based on the received learning model information (M).

Description

Technical Field

[0001] The present invention relates to a diagnosis support system and a diagnosis support device, and in particular to a diagnosis support system and a diagnosis support device having a learning model generation unit that generates learning model information through machine learning based on biological information of a patient group. Background Art

[0002] In the past, there is a known diagnosis support device having a learning model generation unit that generates learning model information by machine learning based on biological information of a patient group. Such a diagnosis support device is disclosed in Japanese Patent Application Publication No. 2018-41434, for example.

[0003] Japanese Patent Publication No. 2018-41434 discloses a diagnostic assistance device for diagnosing lesions based on camera images. In the diagnostic assistance device, it is configured to perform machine learning (neural network) through an integrated identifier (learning model generation unit). Specifically, a plurality of learning images (reference images) with known lesions for machine learning are prepared. Next, a specified image is taken out from the plurality of learning images, and a plurality of images with different rotation angles or magnifications of the taken images are prepared. Then, these images are input into the integrated identifier (neural network) for machine learning. As a result, an integrated identifier that has completed learning is generated. Then, an image with unknown lesions is input into the integrated identifier that has completed learning to infer (determine) whether it is a lesion.

[0004] Prior art literature

[0005] Patent Literature

[0006] Patent Document 1: Japanese Patent Application Publication No. 2018-41434 Summary of the invention

[0007] Problem that the invention aims to solve

[0008] Here, in machine learning such as neural networks as described in Japanese Patent Gazette No. 2018-41434, a relatively large number of images (machine learning data) for machine learning are required. For example, in a relatively large hospital (large hospital), many patients come to the hospital, so it is possible to obtain machine learning data sufficient for machine learning. On the other hand, in a relatively small hospital (small hospital), the number of patients coming to the hospital is small, so it is difficult to obtain machine learning data sufficient for machine learning. In addition, the machine learning data includes personal information for identifying patients, etc., and the machine learning data maintained by large hospitals cannot be used in small hospitals. Therefore, there are the following problems: it is difficult to perform machine learning to determine whether there is a lesion in a small hospital, and it is difficult to make an inference (determination) on whether there is a lesion based on the results of machine learning in a small hospital.

[0009] The present invention is made to solve the above-mentioned problems. One purpose of the present invention is to provide a diagnostic assistance system and a diagnostic assistance device that can determine whether a patient has a disease based on the results of machine learning without leaking the patient's personal information to the outside (small hospitals, etc.) even in small hospitals where it is difficult to obtain data for machine learning.

[0010] Solutions for solving problems

[0011] In order to achieve the above-mentioned purpose, a diagnosis support system according to a first mode of the present invention comprises: a storage unit, which stores biological information of a patient group; a learning model generating unit, which generates a pattern included in the biological information of the patient group, namely, learning model information, by machine learning based on the biological information of the patient group stored in the storage unit; and a determination unit, which receives the learning model information generated by the learning model generating unit via an external network, and determines whether a patient not included in the patient group has a disease based on the received learning model information.

[0012] In the diagnosis support system according to the first mode of the present invention, as described above, there is a determination unit, which receives the learning model information generated by the learning model generation unit via an external network, and determines whether a patient not included in the patient group has a disease based on the received learning model information. Thus, the determination unit set in a small hospital or the like can receive the learning model information via an external network, so even in a small hospital or the like where it is difficult to obtain machine learning data, it is possible to determine whether a patient has a disease based on the learning model information. In addition, the learning model information generated by the learning model generation unit is composed of statistical information that does not include the patient's personal information. Thus, even if the learning model information is provided to the outside (small hospital, etc.) via an external network, the patient's personal information will not be leaked. As a result, even in a small hospital or the like where it is difficult to obtain machine learning data, it is possible to determine whether a patient has a disease based on the result of machine learning (learning model information) without leaking the patient's personal information to the outside (small hospital, etc.).

[0013] In the diagnosis support system according to the first embodiment, preferably, electronic medical record data is stored in the storage unit, the electronic medical record data recording identification information of each patient included in the patient group and biological information of each patient, and the learning model generation unit is configured to extract the biological information of each patient from the electronic medical record data stored in the storage unit, and generate the learning model information based on the extracted biological information of each patient. According to such a configuration, the learning model information is generated based only on the biological information of each patient extracted from the electronic medical record data without using the identification information of the patient, so that the personal information of the patient can be reliably prevented from being leaked.

[0014] In this case, it is preferred that the learning model generation unit is configured to generate the learning model information based on the biological information of each patient extracted from the electronic medical record data stored in the storage unit and the analysis information of the biological sample of each patient associated with the electronic medical record data. According to such a configuration, the learning model information is generated based on the analysis information of the biological sample of each patient in addition to the biological information of each patient, so that it is possible to more accurately determine whether the patient has a disease.

[0015] In the diagnosis support system according to the first embodiment, the determination unit preferably includes a first learning model updating unit, which updates the learning model information received via the external network based on the biological information of patients known to have a disease but not included in the patient group. According to such a configuration, even when it is difficult to obtain data for machine learning, the quality (determination capability) of the learning model information can be improved based on the biological information of patients known to have a disease but not included in the patient group.

[0016] In the diagnosis support system according to the first mode described above, it is preferred that a sending unit is further provided, which sends the biological information of the patient known to have a disease but not included in the patient group to the learning model generation unit via an external network in a state where the patient's identification information is removed, and the learning model generation unit includes a second learning model updating unit, which updates the learning model information based on the biological information sent from the sending unit. According to such a configuration, the quality (determination capability) of the learning model information can be improved by the second learning model updating unit based on the biological information of the patient known to have a disease but not included in the patient group sent from the outside (small hospital, etc.) via an external network. As a result, multiple external institutions such as small hospitals receive the learning model information with improved quality (determination capability) again via the external network, thereby determining whether the patient has a disease based on the learning model information with improved quality and common to multiple external institutions. In addition, the sending unit sends the biological information of the patient to the second learning model updating unit in a state where the patient's identification information is removed, so that the patient's identification information will not be leaked to the outside.

[0017] In the diagnosis support system according to the first aspect, preferably, the learning model information is configured to be exported from the learning model generation unit via an external network and imported into the determination unit. According to such a configuration, even when the application of the learning model generation unit is different from the application of the determination unit, the learning model information is output (exported) from the learning model generation unit in a form read by the application of the determination unit, thereby enabling the learning model information to be used in the determination unit.

[0018] In the diagnosis support system according to the first embodiment, preferably, the biological information of the patient group includes data on the presence or absence of liver cancer, HCV antibody, HBs antigen, age, sex, height, weight, albumin, total bilirubin, AST, ALT, ALP, GGT, platelets, AFP, L3 separation, and DCP, and the determination unit is configured to determine whether a patient not included in the patient group has liver cancer based on the learning model information received via the external network. According to such a configuration, it is possible to determine whether a patient has liver cancer with a relatively high correct answer rate. In addition, it has been confirmed by experiments conducted by the inventors described later that it is possible to determine whether a patient has liver cancer with a relatively high correct answer rate based on the above-mentioned biological information.

[0019] According to the second embodiment of the present invention, the diagnosis assisting device is configured to include: a storage unit, which stores biological information of a patient group; and a learning model generating unit, which generates a pattern included in the biological information of the patient group, i.e., learning model information, through machine learning based on the biological information of the patient group stored in the storage unit. The diagnosis assisting device is configured to export the learning model information generated by the learning model generating unit to the outside via an external network.

[0020] In the diagnostic assisting device according to the second mode of the present invention, as described above, it is configured to export the learning model information generated by the learning model generating unit to the outside via an external network. Thus, since the determination unit provided in a small hospital or the like can receive the learning model information via an external network, even in a small hospital or the like where it is difficult to obtain data for machine learning, it is possible to determine whether a patient has a disease based on the learning model information. In addition, the learning model information generated by the learning model generating unit is composed of statistical information that does not include the patient's personal information. Thus, even if the learning model information is provided to the outside (small hospital or the like) via an external network, the patient's personal information will not be leaked. As a result, it is possible to provide a diagnostic assisting device that can determine whether a patient has a disease based on the results of machine learning (learning model information) without leaking the patient's personal information to the outside (small hospital or the like), even in a small hospital or the like where it is difficult to obtain data for machine learning.

[0021] In the diagnosis support system according to the second embodiment, preferably, electronic medical record data is stored in the storage unit, the electronic medical record data recording identification information of each patient included in the patient group and biological information of each patient, and the learning model generation unit is configured to extract the biological information of each patient from the electronic medical record data stored in the storage unit, and generate the learning model information based on the extracted biological information of each patient. According to such a configuration, the learning model information is generated based only on the biological information of each patient extracted from the electronic medical record data without using the identification information of the patient, so that the personal information of the patient can be reliably prevented from being leaked.

[0022] In this case, it is preferred that the learning model generation unit is configured to generate the learning model information based on the biological information of each patient extracted from the electronic medical record data stored in the storage unit and the analysis information of the biological sample of each patient associated with the electronic medical record data. According to such a configuration, the learning model information is generated based on the analysis information of the biological sample of each patient in addition to the biological information of each patient, so it is possible to more accurately determine whether the patient has a disease.

[0023] Effects of the Invention

[0024] According to the present invention, as described above, even in a small hospital where it is difficult to obtain data for machine learning, it is possible to determine whether a patient has a disease based on the results of machine learning without leaking the patient's personal information to the outside (small hospital, etc.). BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 is a block diagram of a diagnosis support system according to a first embodiment of the present invention.

[0026] Figure 2It is a diagram for explaining the diagnosis support system according to the first embodiment of the present invention.

[0027] Figure 3 is a block diagram of a diagnosis support system according to a second embodiment of the present invention.

[0028] Figure 4 is a block diagram of a diagnosis support system according to a third embodiment of the present invention.

[0029] Figure 5 It is a diagram for explaining a diagnosis support system according to a third embodiment of the present invention.

[0030] Figure 6 is a block diagram of a diagnosis support system according to a fourth embodiment of the present invention. DETAILED DESCRIPTION

[0031] Hereinafter, embodiments of the present invention will be described based on the drawings.

[0032] [First embodiment]

[0033] Reference Figure 1 and Figure 2 The configuration of the diagnosis support system 100 according to the first embodiment will be described.

[0034] First, the electronic medical record (electronic medical record data) is explained. The electronic medical record (electronic medical record data) is data (information system) used to electronically store the doctor's diagnosis record instead of paper. The electronic medical record can be used to improve the efficiency of medical practitioners' office work and integrate information management. In addition, the results obtained in the examination facility where the patient P1 was examined can be automatically associated with the electronic medical record. In addition, in the electronic medical record, the visibility of the text is good, and the electronic medical record can also be easily retrieved.

[0035] In addition, in the diagnosis using previous medical records (electronic medical records, etc.), the doctor makes a comprehensive judgment on the condition of the patient P1 based on the information obtained through the electronic medical records or the medical interview, and makes a diagnosis. In addition, this diagnosis means performing a more invasive and burdensome examination on the patient P1 and determining a treatment policy. In addition, the doctor's comprehensive judgment is based on the doctor's experience supported by statistical insights. On the other hand, the diagnosis support system 100 of the first embodiment performs a diagnosis (supported diagnosis) based on the pattern (learning model information M) included in the biological information of the patient group PF.

[0036] like Figure 1As shown, the diagnosis support system 100 includes: an electronic medical record database 10, a learning model generating unit 11, an electronic medical record database 20, and a determination unit 21. The electronic medical record database 10 and the learning model generating unit 11 are, for example, configured in a facility 1 such as a large hospital where a relatively large number of patients visit the hospital. In addition, the electronic medical record database 20 and the determination unit 21 are, for example, configured in a facility 2 such as a small hospital where a relatively small number of patients visit the hospital. In addition, the electronic medical record database 10 and the learning model generating unit 11 are provided in a diagnosis support device 100a. In addition, the learning model generating unit 11 and the determination unit 21 are composed of software (program). In addition, the electronic medical record database 10 is an example of a "storage unit" in the claims.

[0037] In the electronic medical record database 10, biological information of the patient group PF is stored. Specifically, electronic medical record data is stored in the electronic medical record database 10, and the electronic medical record data records the identification information (name, etc.) of each patient P1 included in the patient group PF and the biological information of each patient P1. Here, the biological information of the patient group PF includes data on the presence or absence of liver cancer, HCV antibodies, HBs antigens, age, gender, height, weight, albumin, total bilirubin, AST, ALT, ALP, GGT, platelets, AFP, L3 separation and DCP. HCV antibodies are indicators indicating whether the patient has been infected with the hepatitis C virus or is currently continuously infected with the hepatitis C virus. HBs antigen is an indicator indicating the current presence of hepatitis B virus (infected). Albumin is a numerical value that measures the concentration of protein in serum, and abnormalities of the liver or kidneys can be investigated based on the reduction of albumin. Total bilirubin is an indicator of the metabolic capacity of the liver. AST is an indicator mainly used to know the degree of obstruction in the liver or heart. ALT is an indicator for knowing whether there is a disorder in the liver. GGT is an indicator of liver function. AFP is an indicator of whether there is liver cancer. L3 separation is a value indicating how much AFP-L3 is included in AFP. DCP is an abnormal prothrombin synthesized by the liver that has no coagulation activity and is a special tumor marker in hepatocellular carcinoma.

[0038] In addition, the learning model generation unit 11 is configured to generate the learning model information M, which is a pattern included in the biological information of the patient group PF, by using machine learning based on the biological information of the patient group PF stored in the electronic medical record database 10. Specifically, in the first embodiment, the learning model generation unit 11 is configured to extract the biological information of each patient P1 from the electronic medical record data stored in the electronic medical record database 10, and generate the learning model information M based on the extracted biological information of each patient P1.

[0039] Here, machine learning refers to repeatedly learning the data that serves as a teacher (data with known judgment results) to find the patterns hidden in the data that serves as a teacher. In machine learning, various algorithms are used to repeatedly learn the data that serves as a teacher, so even if the place that the programmer should explore (the part of the data that serves as a teacher) is not explicitly stated, the computer autonomously derives the pattern. Moreover, in this application specification, the pattern found by machine learning is referred to as learning model information M. When certain data (in the first embodiment, the electronic medical record data of the facility 2 described later) is applied (input) to the learning model information M, the presence or absence of liver cancer is determined based on the learned pattern.

[0040] In addition, the electronic medical record data contains personal information such as the name of the patient P1. In addition, the electronic medical record data contains biological information (data on the presence or absence of liver cancer, HCV antibodies, HBs antigen, age, gender, height, weight, albumin, total bilirubin, AST, ALT, ALP, GGT, platelets, AFP, L3 separation, and DCP). Then, the learning model generation unit 11 extracts the biological information (data on the presence or absence of liver cancer, HCV antibodies, HBs antigen, age, gender, height, weight, albumin, total bilirubin, AST, ALT, ALP, GGT, platelets, AFP, L3 separation, and DCP) from the electronic medical record data.

[0041] Then, the learning model generation unit 11 generates the learning model information M using machine learning such as logistic regression for generating linear learning model information M, soft margin support vector machine (Soft Margin Support Vector Machine), neural network, and random forest for generating nonlinear learning model information M. The generated learning model information M is data having properties close to numerical information after statistical processing and does not include personal information of the patient P1.

[0042] In logistic regression, the variable to be predicted is called the target variable (in the first embodiment, the presence or absence of liver cancer). In addition, the variable that affects the target variable is called the explanatory variable (in the first embodiment, biological information). In logistic regression, the relationship between the target variable and the explanatory variable is represented by a relational expression. In logistic regression, the above-mentioned relational expression is used to calculate the predicted value (the predicted value of the presence or absence of liver cancer) and find the contribution of the explanatory variable to the target variable used in the relational expression.

[0043] In addition, when the support vector machine is provided with teacher data (biological information in the first embodiment), it searches for a hyperplane that separates several biological information (feature quantities) of liver cancer and several biological information (feature quantities) of no liver cancer. In addition, the support vector machine searches for a hyperplane with the largest margin among the multiple hyperplanes that separate the biological information. Here, the margin means the minimum value of the distance between the hyperplane and each feature point, and the hyperplane is searched in a manner that maximizes the margin. In addition, the method of searching for a hyperplane that completely separates the feature points of liver cancer and the feature points of no liver cancer is called a hard margin support vector machine (Japanese: ハードマージンサポートベクターマシン), and the method of searching for a hyperplane in a manner that allows for erroneous judgments on the presence or absence of liver cancer is called a soft margin support vector machine.

[0044] A neural network is a network that represents the neural cells (neurons) and their connections in the human brain as a mathematical model. A neural network consists of an input layer, an output layer, and a hidden layer. Moreover, a weight representing the strength of the connection between neurons is set between each phase. In the learning of the neural network, the weight is adjusted using data that is known to determine whether there is liver cancer, so that the output layer can correctly determine whether there is liver cancer.

[0045] Random forest is an ensemble learning algorithm that combines multiple decision trees. In addition, a decision tree is a tree that finds explanatory variables that affect the target variable and creates a model in the form of a tree.

[0046] like Figure 2 As shown, the learning model generation unit 11 generates learning model information M that distinguishes the biological information of the patient P1 (patient group PF) with liver cancer from the biological information of the patient P1 (patient group PF) without liver cancer. Figure 2 In the above example, the biological information (I1) located above the learning model information M is the biological information of the patient P1 with liver cancer, and the biological information (I2) located below the learning model information M is the biological information of the patient P1 without liver cancer. In addition, the software (application) of machine learning is exemplified in a programming language such as R language. Then, the learning model information M is composed of an object of R language. The object does not include the personal information of the patient P1.

[0047] The diagnosis support device 100a is configured to derive the learning model information M generated by the learning model generation unit 11 from the learning model generation unit 11 via the external network 30. That is, the diagnosis support device 100a is configured to output the learning model information M in a format readable by the determination unit 21.

[0048] Here, in the first embodiment, the determination unit 21 receives the learning model information M generated by the learning model generation unit 11 via the external network 30. Then, the determination unit 21 is configured to determine whether the patient P2 not included in the patient group PF has a disease based on the received learning model information M. In addition, the learning model information M is configured to be exported from the learning model generation unit 11 via the external network 30 and imported into the determination unit 21.

[0049] Specifically, the electronic medical record data of patient P2 is stored in the electronic medical record database 20 of facility 2. In addition, patient P2 is a patient P2 that is not included in the patient P1 (patient group PF) having biological information used when generating the learning model information M. In addition, the electronic medical record data of patient P2 includes biological information (HCV antibody, HBs antigen, age, sex, height, weight, albumin, total bilirubin, AST, ALT, ALP, GGT, platelets, AFP, L3 separation and DCP data) stored in the electronic medical record database 10 of the above-mentioned facility 1. In addition, the electronic medical record data of patient P2 does not include the presence or absence of liver cancer.

[0050] Furthermore, the determination unit 21 is configured to determine whether the patient P2 not included in the patient group PF has liver cancer based on the learning model information M received via the external network 30. Specifically, the biological information included in the electronic medical record data of the patient P2 is imported into the learning model information M. Figure 2 In the example, the presence or absence of liver cancer is determined based on whether the biological information (I3) of the patient P2 is classified to the upper side or the lower side of the learning model information M.

[0051] (experiment)

[0052] Next, an experiment of machine learning based on biological information of patients is described.

[0053] In this experiment, learning model information was generated based on the biological information of 1584 patients who were treated in the outpatient clinic of the hospital. The biological information of the patients is the data of the presence or absence of liver cancer, HCV antibody, HBs antigen, age, gender, height, weight, albumin, total bilirubin, AST, ALT, ALP, GGT, platelets, AFP, L3 separation and DCP. In addition, as machine learning algorithms, logistic regression, soft margin support vector machine, neural network and random forest are used. In addition, cross-validation is performed to determine the correct answer rate of machine learning. In addition, cross-validation refers to dividing the biological information of 1582 patients, generating learning model information through a part of it, and calculating the correct answer rate through the remaining part. As a result, in any machine learning algorithm, it is confirmed that a correct answer rate of about 80% or more is obtained. Therefore, by selecting a machine learning algorithm with, for example, more than 80%, it is possible to use the learning model information M with relatively high accuracy to determine the presence or absence of liver cancer.

[0054] (Effects of the First Embodiment)

[0055] In the first embodiment, the following effects can be obtained.

[0056] In the first embodiment, as described above, the diagnosis support system 100 includes a determination unit 21 that receives the learning model information M generated by the learning model generation unit 11 via the external network 30, and determines whether the patient P2 not included in the patient group PF has a disease (liver cancer) based on the received learning model information M. Thus, since the determination unit 21 provided in a small hospital or the like can receive the learning model information M via the external network 30, it is possible to determine whether the patient P2 has a disease based on the learning model information M even in a small hospital (facility 2) or the like where it is difficult to obtain machine learning data. In addition, the learning model information M generated by the learning model generation unit 11 is composed of statistical information or the like that does not include the personal information of the patient P1. Thus, even if the learning model information M is provided to the outside (small hospital or the like) via the external network 30, the personal information of the patient P1 will not be leaked. As a result, even in a small hospital or the like where it is difficult to obtain machine learning data, it is possible to determine whether the patient P2 has a disease based on the result of machine learning (learning model information M) without leaking the personal information of the patient P1 to the outside (small hospital or the like).

[0057] In the first embodiment, as described above, the learning model generation unit 11 extracts the biological information of each patient P1 from the electronic medical record data stored in the electronic medical record database 10, and generates the learning model information M based on the extracted biological information of each patient P1. Thus, the learning model information M is generated based only on the biological information of each patient P1 extracted from the electronic medical record data without using the identification information of the patient P1, and therefore, it is possible to reliably prevent the personal information of the patient P1 from being leaked.

[0058] In the first embodiment, as described above, the learning model information M is configured to be exported from the learning model generation unit 11 via the external network 30 and imported into the determination unit 21. Thus, even when the application of the learning model generation unit 11 and the application of the determination unit 21 are different, the learning model information M is output (output) from the learning model generation unit 11 in a form read by the application of the determination unit 21, so that the learning model information M can be used in the determination unit 21.

[0059] In the first embodiment, as described above, the biological information of the patient group PF includes data on the presence or absence of liver cancer, HCV antibody, HBs antigen, age, sex, height, weight, albumin, total bilirubin, AST, ALT, ALP, GGT, platelets, AFP, L3 separation, and DCP, and the determination unit 21 is configured to determine whether the patient P2 not included in the patient group PF has liver cancer based on the learning model information M received via the external network 30. Thus, as described in the above-mentioned experiment, it is possible to determine whether the patient P2 has liver cancer at a relatively high correct answer rate (a correct answer rate of approximately 80% or more).

[0060] [Second Embodiment]

[0061] Reference Figure 3 The configuration of the diagnosis support system 200 according to the second embodiment will be described. In the second embodiment, the learning model information M1 is generated based on the analysis information of the biological sample of the patient P1.

[0062] In the diagnosis support system 200 (diagnosis support device 200a), an analysis device 201 for analyzing a biological sample of a patient P1 is provided in the facility 1. The analysis device 201 is, for example, a mass spectrometer. In addition, the analysis device 201 is configured, for example, to identify a molecule that is a marker of a disease (liver cancer) of the patient P1. In addition, the analysis information of the biological sample of the analyzed patient P1 is automatically associated (automatically connected) with the electronic medical record data.

[0063] Then, in the second embodiment, the learning model generation unit 211 is configured to generate the learning model information M1 based on the biological information of each patient P1 extracted from the electronic medical record data stored in the electronic medical record database 10 and the analysis information of the biological sample of each patient P1 associated with the electronic medical record data. That is, the learning model information M1 reflects the biological information of the patient P1 and the analysis information of the biological sample, and in the machine learning of the learning model generation unit 211, the amount of information (feature amount) of the data serving as a teacher becomes larger than that in the first embodiment described above.

[0064] In addition, an analysis device 201 for analyzing a biological sample of patient P2 is also provided in facility 2. The analysis information of the biological sample of patient P2 analyzed by the analysis device 201 is associated with the electronic medical record data of patient P2. Then, the determination unit 221 determines whether patient P2, which is not included in the patient group PF, has a disease based on the learning model information M1 received via the external network 30. Specifically, the biological information included in the electronic medical record data of patient P2 and the analysis information associated with the electronic medical record data are applied (input) to the learning model information M1. Thus, the presence or absence of liver cancer is determined.

[0065] (Effects of the Second Embodiment)

[0066] In the second embodiment, the following effects can be obtained.

[0067] In the second embodiment, as described above, the learning model generation unit 211 generates the learning model information M1 based on the biological information of each patient P1 extracted from the electronic medical record data stored in the electronic medical record database 10 and the analysis information of the biological sample of each patient P1 associated with the electronic medical record data. Thus, the learning model information M is generated based on the analysis information of the biological sample of each patient P1 in addition to the biological information of each patient P1, so it is possible to more accurately determine whether the patient P2 has a disease.

[0068] [Third Embodiment]

[0069] Reference Figure 4 and Figure 5 The configuration of the diagnosis support system 200 according to the third embodiment will be described. In the third embodiment, the learning model information M is updated based on the biological information of the patient P2.

[0070] In the third embodiment, the determination unit 321 of the diagnosis support system 300 includes a learning model updating unit 322. The learning model updating unit 322 is configured to update the learning model information M received via the external network 30 based on the biological information of the patient P2 who is not included in the patient group PF and is known to have a disease. Here, the learning model information M is generated by performing machine learning on all the biological information of the patient group PF. Then, the learning model updating unit 322 updates the learning model information M only based on the biological information of the patient P2 without using the biological information of the patient group PF, thereby generating the learning model information M2. As a result, the biological information of the patient P2 is reflected in the learning model information M2. For example, the parameters of the learning model information M2 are updated based on the biological information of the patient P2. In addition, the learning model updating unit 322 is an example of the "first learning model updating unit" in the claims.

[0071] (Effects of the Third Embodiment)

[0072] In the third embodiment, the following effects can be obtained.

[0073] In the third embodiment, as described above, the determination unit 21 includes a learning model updating unit 322 that updates the learning model information M received via the external network 30 based on the biological information of the patient P2 known to have a disease but not included in the patient group PF. Thus, even when it is difficult to obtain data for machine learning, the quality (determination capability) of the learning model information M2 can be improved based on the biological information of the patient P2 known to have a disease but not included in the patient group PF.

[0074] [Fourth Embodiment]

[0075] Reference Figure 6 The configuration of a diagnosis support system 400 according to the fourth embodiment will be described. In the fourth embodiment, a transmission unit 410 that transmits biological information I3 of a patient P2 is provided.

[0076] In the fourth embodiment, a sending unit 410 is provided in the diagnosis support system 400. The sending unit 410 is provided in the facility 2. The sending unit 410 is configured to send the biological information I3 of the patient P2 whose presence or absence of a disease is known but not included in the patient group PF to the learning model generation unit 411 via the external network 30 in a state where the identification information of the patient P2 is removed. Specifically, the biological information I3 of each patient P2 is extracted from the electronic medical record data stored in the electronic medical record database 20 of the facility 2. In addition, the identification information (personal information such as name) of the patient P2 is not extracted. Then, the sending unit 410 sends the extracted biological information I3 of each patient P2 to the learning model generation unit 411.

[0077] In addition, the learning model generation unit 411 includes a learning model update unit 412. The learning model update unit 412 is configured to update the learning model information M based on the biological information I3 of the patient P2 whose presence or absence of a disease is known but not included in the patient group PF. The learning model update unit 412 updates the learning model information M based only on the biological information I3 of the patient P2 without using the biological information of the patient group PF. In addition, the biological information I3 includes data on the presence or absence of liver cancer, HCV antibody, HBs antigen, age, sex, height, weight, albumin, total bilirubin, AST, ALT, ALP, GGT, platelets, AFP, L3 separation, and DCP. In addition, the sending unit 410 may also periodically send the biological information I3 of the patient P2 to the learning model generation unit 11. Thus, the learning model information M is periodically updated. In addition, the learning model update unit 412 is an example of the "second learning model update unit" in the claims.

[0078] (Effects of the Fourth Embodiment)

[0079] In the fourth embodiment, the following effects can be obtained.

[0080] In the fourth embodiment, as described above, a sending unit 410 is provided, which sends the biological information I3 of the patient P2 whose disease is known but not included in the patient group PF to the learning model generation unit 11 via the external network 30 in a state where the identification information of the patient P2 is removed. As a result, the quality (determination capability) of the learning model information M can be improved by the learning model update unit 412 based on the biological information I3 of the patient P2 whose disease is known but not included in the patient group PF and which is sent from the outside (facility 2) via the external network 30. As a result, the plurality of facilities 2 receive the learning model information M whose quality (determination capability) has been improved again via the external network 30, thereby determining whether the patient (patient whose disease is unknown) has a disease based on the learning model information M whose quality has been improved and which is common in the plurality of facilities 2. In addition, since the sending unit 410 sends the biological information I3 of the patient P2 to the learning model update unit 412 in a state where the identification information of the patient P2 is removed, the identification information of the patient P2 is not leaked to the outside (facility 1, etc.).

[0081] [Modifications]

[0082] The embodiments disclosed this time should be considered in all respects as illustrative and non-restrictive. The scope of the present invention is indicated by the claims rather than the description of the embodiments described above, and includes all changes (modifications) within the meaning and scope equivalent to the claims.

[0083] For example, in the first to fourth embodiments, machine learning is performed on the biological information of each patient extracted from the electronic medical record data, but the present invention is not limited thereto. For example, machine learning may also be performed based on the biological information of each patient other than the electronic medical record data.

[0084] In the first to fourth embodiments, the presence or absence of liver cancer is determined based on the learned model information, but the present invention is not limited thereto and can also be applied to the determination of diseases other than liver cancer (eg, pancreatic cancer).

[0085] In addition, in the first to fourth embodiments, the following example is shown: as the biological information of the patient group, all the data of HCV antibody, HBs antigen, age, sex, height, weight, albumin, total bilirubin, AST, ALT, ALP, GGT, platelets, AFP, L3 separation and DCP are used, but the present invention is not limited to this. For example, as the biological information of the patient group, part of the data of HCV antibody, HBs antigen, age, sex, height, weight, albumin, total bilirubin, AST, ALT, ALP, GGT, platelets, AFP, L3 separation and DCP may be used.

[0086] In addition, in the present application specification, the second embodiment of generating learning model information based on biological information extracted from electronic medical record data and analysis information of biological samples, and the third and fourth embodiments of updating the learning model information are described as different embodiments; however, the structure of the second embodiment, the structure of the third embodiment, and the structure of the fourth embodiment may be combined.

[0087] In addition, in the first to fourth embodiments, the following examples are shown: logistic regression, soft margin support vector machine, neural network and random forest are used as machine learning algorithms, but the present invention is not limited to this. For example, as a machine learning algorithm, a machine learning algorithm other than logistic regression, soft margin support vector machine, neural network and random forest may also be used.

[0088] Description of Reference Numerals

[0089] 10: Electronic medical record database (storage unit); 11, 211, 411: Learning model generation unit; 21, 221, 321: Determination unit; 30: External network; 100, 200, 300, 400: Diagnosis support system; 100a, 200a: Diagnosis support device; 322: Learning model update unit (first learning model update unit); 401: Sending unit; 412: Learning model update unit (second learning model update unit); M, M1, M2: Learning model information; P1, P2: Patient; PF: Patient group.

Claims

1. A diagnostic assistance system comprising: a storage unit, which is disposed in a facility and stores biological information of a patient group; a learning model generating unit that generates learning model information for determining the presence or absence of a disease by machine learning based on biological information of a patient group arranged in the one facility; as well as A determination unit is disposed in a hospital different from the one facility, receives the learning model information generated by the learning model generation unit from a computer disposed outside the hospital via an external network, and determines whether a patient not included in the patient group has a disease by importing biological information of the patient not included in the patient group into the received learning model information.

2. The diagnostic assistance system according to claim 1, wherein: further comprising a storage unit storing the biological information of the patient group, The storage unit stores electronic medical record data, wherein the electronic medical record data records identification information of each patient included in the patient group and the biological information of each patient. The learning model generating unit is configured to extract the biological information of each patient from the electronic medical record data stored in the storage unit, and to generate the learning model information based on the extracted biological information of each patient.

3. The diagnostic assistance system according to claim 2, wherein: The learning model generating unit is configured to generate the learning model information based on the biological information of each patient extracted from the electronic medical record data stored in the storage unit and analysis information of a biological sample of each patient associated with the electronic medical record data.

4. The diagnosis support system according to any one of claims 1 to 3, wherein: The determination unit includes a first learning model updating unit that updates the learning model information received via the external network based on the biological information of patients whose presence or absence of a disease is known and who are not included in the patient group.

5. The diagnostic assistance system according to claim 1 or 2, wherein: The hospital further comprises a sending unit, the sending unit being configured in the hospital and sending the biological information of patients known to have a disease but not included in the patient group to the learning model generating unit via the external network in a state where the identification information of the patients is removed. The learning model generating unit includes a second learning model updating unit that updates the learning model information based on the biological information transmitted from the transmitting unit.

6. The diagnostic assistance system according to claim 1 or 2, wherein: The learning model information is configured to be exported from the learning model generation unit via the external network from the computer disposed outside the hospital, and to be imported into the determination unit disposed in the hospital.

7. The diagnostic assistance system according to claim 1 or 2, wherein: The biological information of the patient group includes data on the presence or absence of liver cancer, HCV antibody, HBs antigen, age, gender, height, weight, albumin, total bilirubin, AST, ALT, ALP, GGT, platelets, AFP, L3 separation and DCP. The determination unit is configured to determine whether a patient not included in the patient group has liver cancer based on the learning model information received via the external network.

8. A diagnostic aid method comprising the following steps: In one facility, a learning model generation unit generates learning model information for determining the presence or absence of a disease by machine learning based on biological information of a patient group; The diagnosis support device exports the generated learning model information to the outside of the one facility via an external network; and The determination unit receives the generated learning model information from a computer configured in a hospital different from the one facility via the external network, and determines whether the patient not included in the patient group has a disease by importing the biological information of the patient not included in the patient group into the received learning model information.

Images