Program, trained model generation method, information processing method, and information processing apparatus

Abstract

Provided is, for example, a program capable of easily outputting effect information relating to treatment effect on rheumatoid arthritis.　The program acquires molecular targeting drug administration information, the type and dosage of a disease-modifying antirheumatic drug, patient information regarding a patient, and oral glucocorticoid administration information, and outputs effect information in response to the input of the molecular targeting drug administration information, the type and dosage of the disease-modifying antirheumatic drug, the patient information regarding the patient, and the oral glucocorticoid administration information into a trained model that outputs effect information relating to treatment effect on rheumatoid arthritis when the molecular targeting drug administration information, the type and dosage of the disease-modifying antirheumatic drug, the patient information regarding the patient, and the oral glucocorticoid administration information are input.

Description

Program, Method for Generating Learning Model, Information Processing Method, and Information Processing Apparatus

[0001] The present disclosure relates to a program, a method for generating a learning model, an information processing method, and an information processing apparatus.

[0002] In Patent Document 1, the expression level of a biomarker gene for predicting the therapeutic effect of a TNF inhibitor, tocilizumab, or abatacept, which are therapeutic agents for rheumatoid arthritis, is measured, and based on the measured expression level, a method of selecting an optimal therapeutic agent from a TNF inhibitor, tocilizumab, or abatacept for each patient is disclosed.

[0003] Japanese Patent No. 6985691

[0004] However, the invention according to Patent Document 1 has a problem that effect information regarding the therapeutic effect of rheumatoid arthritis cannot be easily output.

[0005] In one aspect, an object is to provide a program or the like that can easily output effect information regarding the therapeutic effect of rheumatoid arthritis.

[0006] A program according to one aspect causes a computer to execute a process of outputting effect information by inputting the administration information of a molecular target drug, the type and dosage of a disease-modifying antirheumatic drug, the patient information of a patient, and the administration information of an oral glucocorticoid into a learning model that outputs effect information regarding the therapeutic effect of rheumatoid arthritis when the administration information of the molecular target drug, the type and dosage of the disease-modifying antirheumatic drug, the patient information of the patient, and the administration information of the oral glucocorticoid are input.

[0007] In one aspect, it becomes possible to easily output effect information regarding the therapeutic effect of rheumatoid arthritis.

[0008] This is an explanatory diagram showing an overview of an information processing system. This is a block diagram showing an example of a server configuration. This is a block diagram showing an example of a computer configuration. This is a schematic diagram showing an example of a learning model configuration. This is an explanatory diagram showing an example of a training DB record layout. This is an explanatory diagram showing an example of a screen. This is an explanatory diagram showing an example of a screen. This is an explanatory diagram showing an example of a screen. This is a flowchart showing an example of a learning model generation procedure. This is a flowchart showing an example of a processing procedure when outputting effect information. This is an explanatory diagram showing an example of a screen. This is an explanatory diagram showing an example of a screen. This is a flowchart showing an example of a processing procedure when outputting effect information based on changed input information. This is a flowchart showing an example of a processing procedure related to retraining. This is a flowchart showing a processing procedure when outputting effect information based on changed input information.

[0009] (Embodiment 1) Figure 1 is an explanatory diagram showing an overview of the information processing system. The information processing system outputs efficacy information regarding the therapeutic effect of rheumatoid arthritis based on the administration information of molecularly targeted drugs, the type and dosage of disease-modifying antirheumatic drugs, patient information, and the administration information of oral glucocorticoids. The information processing system includes an information processing device 10 and an information processing device 20. The information processing system transmits and receives information via a network N.

[0010] The information processing device 10 is an information processing device that processes, stores, and transmits various types of information. The information processing device 10 is, for example, a server device, a personal computer, or a general-purpose tablet PC (personal computer). The information processing device 10 may also be a cloud server device that provides the functions included in the information processing device 10 as a cloud service. In this embodiment, the information processing device 10 will be described as a server 10.

[0011] The information processing device 20 is, for example, a server device, a smartphone, a tablet, a personal computer (hereinafter referred to as "computer"), or a general-purpose tablet PC (personal computer). In this embodiment, the information processing device 20 will be described as a computer 20.

[0012] Figure 2 is a block diagram showing an example of a server configuration. The server 10 includes a control unit 11, a storage unit 12, a communication unit 13, a large-capacity storage unit 14, and a read unit 15. Each of the above-mentioned units is interconnected via a bus. The control unit 11 is configured using one or more processors such as a CPU (Central Processing Unit), MPU (Micro-Processing Unit), GPU (Graphics Processing Unit), GPGPU (General-purpose computing on graphics processing units), TPU (Tensor Processing Unit), or an AI chip (AI semiconductor). The control unit 11 performs various information processing and control processing related to the server 10 by appropriately executing a control program 12P (program product) stored in the storage unit 12.

[0013] The storage unit 12 includes RAM (Random Access Memory) or ROM (Read Only Memory), etc. The storage unit 12 stores various data necessary for the control program 12P executed by the control unit 11. The storage unit 12 temporarily stores data generated when the control unit 11 executes the control program 12P.

[0014] The communication unit 13 is a communication module that sends and receives information to and from the computer 20 via the network N. The large-capacity storage unit 14 includes RAM or ROM, etc. The large-capacity storage unit 14 stores the learning model 14M and the training DB 141. The learning model 14M and the training DB 141 will be described later. In this embodiment, the storage unit 12 and the large-capacity storage unit 14 may be configured as an integrated storage device. The large-capacity storage unit 14 may be composed of multiple storage devices. The large-capacity storage unit 14 may also be an external storage device connected to the server 10.

[0015] The reading unit 15 reads information stored in the portable storage medium 1a. The portable storage medium 1a is, for example, a CD (Compact Disc), DVD (Digital Versatile Disc), USB (Universal Serial Bus) memory, or SD (Secure Digital). The reading unit 15 reads the control program 12P from the portable storage medium 1a. The control unit 11 stores the read control program 12P in the storage unit 12. The control unit 11 may also download the control program 12P from another computer via the network N. In that case, the control unit 11 stores the downloaded control program 12P in the storage unit 12. The control unit 11 may also store the read control program 12P in the large-capacity storage unit 14.

[0016] In this embodiment, server 10 may be composed of multiple servers. Server 10 may be a virtual machine virtually constructed by software within a single device. Server 10 may be a local server installed in the facility where server 10 is located. Server 10 may be a cloud server connected via network N. Furthermore, the control program 12P may be executed on a single server, or it may be executed in a distributed manner on multiple servers interconnected via network N.

[0017] Figure 3 is a block diagram showing an example of a computer configuration. The computer 20 includes a control unit 21, a storage unit 22, a communication unit 23, an input unit 24, and a display unit 25. The above-mentioned units are interconnected via a bus. The control unit 21 is configured using one or more processors such as a CPU, MPU, or GPU. The control unit 21 performs various information processing and control processing related to the computer 20 by appropriately executing the control program 22P stored in the storage unit 22. The storage unit 22 includes RAM or ROM, etc. The storage unit 22 stores various data necessary for the control program 22P (program product) executed by the control unit 21. The storage unit 22 temporarily stores data generated when executing the control program 22P.

[0018] The communication unit 23 is a communication module that sends and receives information to and from the server 10 via the network N. The input unit 24 receives input from medical personnel. The input unit 24 includes, for example, a keyboard, mouse, and microphone for voice input. The display unit 25 is a liquid crystal display or an organic EL display, etc. The display unit 25 outputs various types of information according to instructions from the control unit 21. The input unit 24 and the display unit 25 may be integrated into a single touch panel.

[0019] Figure 4 is a schematic diagram showing an example of the configuration of a learning model. When learning model 14M is input with information on the administration of molecularly targeted drugs, the type and dosage of disease-modifying rheumatoid drugs, patient information, and oral glucocorticoid administration information, it outputs effect information regarding the therapeutic effect of rheumatoid arthritis. The effect information (output data) includes, for example, the probability that the therapeutic effect will be effective, whether or not the therapeutic effect is effective, or the level (score) indicating the effectiveness of the therapeutic effect. The probability that the therapeutic effect will be effective is shown as a value from 0 to 1, for example. The closer the probability of the therapeutic effect will be to "0", the less effective the therapeutic effect on rheumatoid arthritis will be, and the closer the probability of the therapeutic effect will be to "1", the more effective the therapeutic effect on rheumatoid arthritis will be.

[0020] Whether or not a treatment is effective is indicated, for example, as "Yes (treatment is effective)" or "No (treatment is not effective)." The level of effectiveness is indicated in several stages, for example, as "Level 1," "Level 2," "Level 3," "Level 4," or "Level 5." A lower level of effectiveness indicates that the treatment is not effective for rheumatoid arthritis, while a higher level indicates that the treatment is effective for rheumatoid arthritis. Hereafter, effectiveness information will be explained as the probability of the treatment being effective.

[0021] Molecularly targeted drugs include, for example, at least one of a biological agent and a JAK inhibitor. Biological agents include, for example, infliximab, tocilizumab, or abatacept. JAK inhibitors include, for example, tofacitinib, baricitinib, peficitinib, or upadacitinib. Information on the administration of molecularly targeted drugs includes at least one of the following: whether or not the molecularly targeted drug was administered, the type of molecularly targeted drug, the dosage and the start date of administration, and the date and reason for discontinuation of the molecularly targeted drug.

[0022] The type of molecularly targeted drug includes at least one of the following: the name of the molecularly targeted drug to be administered, the name of the molecularly targeted drug currently being administered, and the name of the molecularly targeted drug previously administered. The dosage of the molecularly targeted drug includes at least one of the following: the dosage of the molecularly targeted drug to be administered, the dosage of the molecularly targeted drug currently being administered, and the dosage of the molecularly targeted drug previously administered. The start date of administration of the molecularly targeted drug includes at least one of the following: the start date of administration of the molecularly targeted drug to be administered, the start date of administration of the molecularly targeted drug currently being administered, and the start date of administration of the molecularly targeted drug previously administered. The reason for discontinuation of the molecularly targeted drug includes at least one of the following: drug ineffectiveness, adverse event, non-adverse event, and remission.

[0023] Disease-modifying rheumatic drugs include methotrexate (MTX) and at least one disease-modifying antirheumatic drug other than methotrexate. Disease-modifying antirheumatic drugs other than methotrexate include at least one of tacrolimus, iguratimod, bucillamine, sulfasalazine, leflunomide, sodium aurothiomalate, penicillamine, actarit, azathioprine, and mizoribine.

[0024] Patient information includes age, sex, height, weight, duration of illness, anti-CCP antibody titer, rheumatoid factor antibody titer, blood test data, rheumatoid arthritis physical disability assessment index, and at least one rheumatoid arthritis disease activity assessment index. Blood test data included in patient information includes white blood cell count, percentage of each white blood cell type, red blood cell count, platelet count, CRP (C-reactive protein) value, ESR (Erythrocyte Sedimentation Rate), MMP-3 (Matrix Metalloproteinase-3) value, eGFR (Estimated Glomerular Filtration Rate), and at least one KL-6 (krebs von den lungen-6) value.

[0025] The physical disability assessment index for rheumatoid arthritis includes at least one of the HAQ (Health Assessment Questionnaire) and mHAQ (modified Health Assessment Questionnaire). The disease activity assessment index for rheumatoid arthritis includes at least one of the DAS28-CRP (Disease Activity Score using 28 joint counts and C-reactive protein) and CDAI (Clinical Disease Activity Index). In the following explanation, the physical disability assessment index for rheumatoid arthritis will be described as HAQ, and the disease activity assessment index for rheumatoid arthritis will be described as DAS28-CRP.

[0026] Information on oral glucocorticoid administration includes at least one of the following: whether or not oral glucocorticoids were administered, the dosage and start date of oral glucocorticoids planned for administration, the dosage and start date of oral glucocorticoids currently being administered, the dosage and start date of oral glucocorticoids administered, and the date and reason for discontinuation of oral glucocorticoids that were discontinued. The reason for discontinuation of oral glucocorticoids includes at least one of the following: drug ineffectiveness, adverse events, non-adverse events, and remission.

[0027] In this embodiment, an example is shown in which the learning model 14M outputs effect information regarding the therapeutic effect on rheumatoid arthritis when it receives input information on the administration of molecularly targeted drugs, the type and dosage of disease-modifying rheumatoid drugs, patient information, and oral glucocorticoid administration information, but it is not limited to this. The learning model 14M may also output effect information regarding the therapeutic effect on rheumatoid arthritis when it receives input for at least two of the following: information on the administration of molecularly targeted drugs, the type and dosage of disease-modifying rheumatoid drugs, patient information, and oral glucocorticoid administration information.

[0028] The learning model 14M according to this embodiment is composed of a random forest. The learning model 14M may be, for example, a Transformer, BERT (Bidirectional Encoder Representations from Transformers), GPT (Generative Pre-trained Transformer), CNN (Convolutional Neural Network), RNN (Recurrent Neural Network), LSTM (Long Short Term Memory), Support Vector Machine, Logistic Regression, Random Forest, or other learning algorithms such as XGBoost (eXtreme Gradient Boosting). The learning model 14M may be composed of a combination of multiple algorithms.

[0029] The learning model 14M is composed of multiple decision trees, each having branching nodes and leaf nodes at the end of those branching nodes. In each decision tree, input data, including information on the administration of molecularly targeted drugs, the type and dosage of disease-modifying rheumatic drugs, patient information, and oral glucocorticoid administration information, is classified according to conditions as it moves from the top root node to the branches. When it reaches a terminal leaf node, the value (0 to 1) assigned to that terminal leaf node is output as effect information (output data). Each leaf node is associated with a value (0 to 1) of effect information. The learning model 14M is generated using a large amount of training data that links input data, including information on the administration of molecularly targeted drugs, the type and dosage of disease-modifying rheumatic drugs, patient information, and oral glucocorticoid administration information, with the correct value labels (effect information) of the input data.

[0030] Figure 5 is an explanatory diagram showing an example of the record layout of the training database. The training database 141 stores a large amount of training data used to generate the learning model 14M. The training database 141 includes columns for ground truth labels, molecular targeted drugs, disease-modifying rheumatic drugs, patient information, and oral glucocorticoids. The ground truth label column stores information indicating whether the treatment effect was effective or not ("1": treatment effect was effective / "0": treatment effect was not effective).

[0031] The molecular targeted drug column stores information about the administration of molecular targeted drugs and includes columns for administration status, administered drug, dosage, administration start date and time, discontinued drug, discontinued date and time, and reason for discontinuation. The administration status column stores whether or not the molecular targeted drug was administered to the patient. If the molecular targeted drug was administered to the patient, "Yes" is stored in the administration status column; if the molecular targeted drug was not administered, "No" is stored in the administration status column. If the molecular targeted drug was not administered, nothing is stored in the administered drug, dosage, administration start date and time, discontinued drug, discontinued date and time, and reason for discontinuation columns. The following explanation will use an example where the molecular targeted drug was administered to the patient.

[0032] The "Administered Drug" column stores the type of molecularly targeted drug administered. The "Dosage" column stores the dosage for each molecularly targeted drug administered. The "Administration Start Date and Time" column stores the administration start date and time for each molecularly targeted drug administered. The "Discontinued Drug" column stores the type of molecularly targeted drug that was discontinued. The "Discontinuation Date and Time" column stores the discontinuation date and time for each molecularly targeted drug that was discontinued. The "Discontinuation Reason" column stores the reason for discontinuation for each molecularly targeted drug that was discontinued. If there is no information to be entered in one or more of the "Administered Drug" column, "Dosage" column, "Administration Start Date and Time" column, "Discontinued Drug" column, "Discontinuation Date and Time" column, and "Discontinuation Reason" column, then "- (Not Entered)" may be stored in each column.

[0033] The disease-modifying rheumatoid arthritis drug column stores the type and dosage of the administered disease-modifying rheumatoid arthritis drug, and includes a drug column and a dosage column. The drug column stores the type of disease-modifying rheumatoid arthritis drug administered. The dosage column stores the dosage for each administered disease-modifying rheumatoid arthritis drug.

[0034] The patient information column stores information about patients receiving treatment for rheumatoid arthritis and includes columns for gender, age, height, weight, duration of illness, anti-CCP antibody, rheumatoid factor, blood test data, physical disability assessment index, and disease activity assessment index. The gender column stores the patient's gender. The age column stores the patient's age. The height column stores the patient's height. The weight column stores the patient's weight. The duration of illness column stores the length of time the patient has had rheumatoid arthritis. The anti-CCP antibody column stores the patient's anti-CCP antibody titer. The rheumatoid factor column stores the patient's rheumatoid factor antibody titer.

[0035] The blood sampling data column stores the patient's blood sampling data. The physical disability assessment index column stores HAQ, an example of a physical disability assessment index for rheumatoid arthritis. The disease activity assessment index column stores DAS28-CRP, an example of a disease activity assessment index for rheumatoid arthritis. If there is no information to be entered in one or more of the gender, age, height, weight, duration of illness, anti-CCP antibody, rheumatoid factor, blood sampling data, physical disability assessment index, and disease activity assessment index columns, then "- (not entered)" may be stored in each column.

[0036] The oral glucocorticoid column stores information on the administration of oral glucocorticoids and includes columns for dosage, administration start date and time, discontinuation date and time, and discontinuation reason. The dosage column stores the dosage of oral glucocorticoids in prednisolone equivalent. The administration start date and time column stores the start date and time of administration of oral glucocorticoids. The discontinuation date and time column stores the date and time of discontinuation if oral glucocorticoid administration is discontinued. The discontinuation reason column stores the reason for discontinuation if oral glucocorticoid administration is discontinued. If there is no information to be entered in one or more of the dosage, administration start date and time, discontinuation date and time, and discontinuation reason columns, "- (not entered)" may be stored in each column.

[0037] The learning model 14M is trained by rearranging the order of conditions at each node so that when input data including information on the administration of molecularly targeted drugs, the type and dosage of disease-modifying rheumatic drugs, patient information, and information on the administration of oral glucocorticoids is input, it outputs the same value (for example, 1 or 0) as the correct label corresponding to the input data. Specifically, if the correct label corresponding to the input data is "0 (treatment effect is not effective)", the learning model 14M is trained to output "0 (treatment effect is not effective)" when that input data is input. The control unit 11 generates (constructs) the learning model by performing similar learning using a large amount of training data.

[0038] In this embodiment, an example is shown in which a learning model 14M is generated using all the information stored in the molecular targeted drug list, disease-modifying rheumatoid drug list, patient information list, and oral glucocorticoid list as training data, but the embodiment is not limited to this. The learning model 14M may be generated using any or more of the information stored in the molecular targeted drug list, disease-modifying rheumatoid drug list, patient information list, and oral glucocorticoid list as training data.

[0039] The learning model 14M may be generated on an external device (not shown) other than the server 10. In that case, the generated learning model 14M is deployed to the server 10 from an external computer or the like via the network N. Alternatively, the generated learning model 14M may be deployed to computer 20. In that case, computer 20 outputs effect information on the input data.

[0040] The process by which the information processing system outputs effect information using the learning model 14M will be described.When a healthcare professional executes the program of this embodiment, the control unit 21 displays a screen on the display unit 25 for receiving input of molecular targeted drug administration information, disease-modifying rheumatic drug type and dosage, patient information, and oral glucocorticoid administration information.

[0041] Figures 6 and 7 are explanatory diagrams showing an example of a screen. Screen d100 in Figure 6 is the first screen displayed on the display unit 25 when a healthcare professional executes the program of this embodiment. Screen d100 in Figure 6 includes input fields, pull-down lists, slide switches, and checkboxes for inputting or selecting information on the administration of molecularly targeted drugs, the type and dosage of disease-modifying rheumatoid arthritis drugs, and patient information.

[0042] The screen d100 in Figure 7 is the screen displayed on the display unit 25 when a healthcare professional scrolls the screen d100 in Figure 6 upwards. The screen d100 shown in Figure 7 includes input fields for inputting or selecting patient information and oral glucocorticoid administration information, as well as a slide switch and a pull-down list.

[0043] At the bottommost row of the screen d100, a transmission button b1 is arranged. The transmission button b1 is a button for transmitting to the server 10 the administration information of the molecular target drug, the type and dosage of the disease-modifying antirheumatic drug, patient information, and the administration information of oral glucocorticoid, which have received an input or selection. The control unit 21 receives the input or selection of the administration information of the molecular target drug, the type and dosage of the disease-modifying antirheumatic drug, the patient information of the patient, and the administration information of oral glucocorticoid via an input field, a slide switch, a pull-down list, or a check box on the screen d100.

[0044] The control unit 21 determines whether or not the selection of the transmission button b1 has been received. When the control unit 21 determines that the selection of the transmission button b1 has not been received, it receives again the input or selection of the administration information of the molecular target drug, the type and dosage of the disease-modifying antirheumatic drug, the patient information of the patient, and the administration information of oral glucocorticoid. When the control unit 21 determines that the selection of the transmission button has been received, it transmits to the server 10 the administration information of the molecular target drug, the type and dosage of the disease-modifying antirheumatic drug, patient information, and the administration information of oral glucocorticoid, which have received an input or selection, via the communication unit 23.

[0045] Embodiment 1 shows an example in which the control unit 21 acquires the administration information of the molecular target drug, the type and dosage of the disease-modifying antirheumatic drug, patient information, and the administration information of oral glucocorticoid via the screen d100 shown in FIGS. 6 and 7, but is not limited thereto. The control unit 21 may acquire the administration information of the molecular target drug, the type and dosage of the disease-modifying antirheumatic drug, patient information, and the administration information of oral glucocorticoid from the content stored in an electronic medical record or the like.

[0046] The communication unit 13 receives the administration information of the molecular target drug, the type and dosage of the disease-modifying antirheumatic drug, patient information, and the administration information of oral glucocorticoid transmitted from the computer 20. When there is a missing part in any of the received administration information of the molecular target drug, the type and dosage of the disease-modifying antirheumatic drug, patient information, and the administration information of oral glucocorticoid, the control unit 11 complements the missing part by the multiple complement method.

[0047] The control unit 11 obtains the effect information output by the learning model 14M by inputting the administered information of the received molecular target drug, the type and dosage of the disease-modifying antirheumatic drug, the patient information, and the administered information of the oral glucocorticoid into the learning model 14M. The control unit 11 stores the obtained effect information in the mass storage unit 14 in association with the administered information of the molecular target drug, the type and dosage of the disease-modifying antirheumatic drug, the patient information, and the administered information of the oral glucocorticoid. The communication unit 13 transmits the administered information of the molecular target drug, the type and dosage of the disease-modifying antirheumatic drug, the patient information, and the administered information of the oral glucocorticoid, and the effect information to the computer 20. The communication unit 13 may transmit only the obtained effect information to the computer 20.

[0048] The communication unit 23 receives the administered information of the molecular target drug, the type and dosage of the disease-modifying antirheumatic drug, the patient information, and the administered information of the oral glucocorticoid, and the effect information transmitted from the server 10. The control unit 21 displays the received administered information of the molecular target drug, the type and dosage of the disease-modifying antirheumatic drug, the patient information, and the administered information of the oral glucocorticoid, and the effect information on the display unit 25.

[0049] FIG. 8 is an explanatory diagram showing an example of a screen example. The screen d200 includes an effect information column d21 and an input data column d22. In the effect information column d21, the effect information (probability that the treatment effect is effective) output by the learning model 14M is displayed. In the effect information column d21 of FIG. 8, as an example of the effect information, "probability that the treatment effect is effective: 0.8" is displayed. In the input data column d22, the administered information of the molecular target drug, the type and dosage of the disease-modifying antirheumatic drug, the patient information, and the administered information of the oral glucocorticoid input to the learning model are displayed.

[0050] In the input data column d22 of FIG. 8, the administered information of the molecular target drug, the type and dosage of the disease-modifying antirheumatic drug, the patient information, and the administered information of the oral glucocorticoid are displayed for each item. In the present embodiment, an example in which the screen d200 includes the effect information column d21 and the input data column d22 is shown, but it is not limited thereto. The screen d200 may include only the effect information column d21.

[0051] (Modification 1) Embodiment 1 shows an example in which patient information includes age, sex, height, weight, duration of illness, antibody titer of anti-CCP antibody, antibody titer of rheumatoid factor, blood test data, rheumatoid arthritis physical disability assessment index, and rheumatoid arthritis disease activity assessment index, but is not limited to this. In addition to the above information, the patient information may further include at least one of the number of tender joints and the number of swollen joints, the site corresponding to the number of tender joints or swollen joints, and information on orthopedic surgery.

[0052] Let's explain that case. The number of tender joints is the number of joints among the DAS28 joints that are tender to the touch (hereinafter referred to as tender joints), and is shown, for example, as 1 to 28. The number of swollen joints is the number of joints among the DAS28 joints that show swelling (hereinafter referred to as swollen joints), and is shown, for example, as 1 to 28. The joints counted as tender joints and swollen joints may include joints other than the DAS28 joints (for example, ankle joints).

[0053] The areas corresponding to the number of tender joints are the joints that are counted as tender, and are indicated by, for example, the "right shoulder," "left wrist," and the "metacarpophalangeal joints (MCP) of the left index finger." The areas corresponding to the number of swollen joints are the joints that are counted as swollen joints, and are indicated by, for example, the "left shoulder," "left wrist," and the "proximal interphalangeal joints (PIP) of the right thumb." Areas corresponding to the number of tender joints and areas corresponding to the number of swollen joints may overlap.

[0054] Information regarding orthopedic surgery includes, for example, the type of orthopedic surgery to be performed, the type of orthopedic surgery currently being performed, and at least one of the types of orthopedic surgery performed in the past. The types of orthopedic surgery include at least one of the following: artificial joint replacement, arthroplasty, arthrodesis, synovectomy, and spinal surgery (e.g., spinal fusion).

[0055] The patient information column in the training data DB141 (see Figure 5) further includes a column that stores at least one of the number of tender joints and the number of swollen joints, a column that stores at least one of the body parts corresponding to the number of tender joints or swollen joints, and a column that stores information related to orthopedic surgery. The control unit 11 generates (constructs) a learning model 14M using a large number of training data that further include as patient information at least one of the number of tender joints and the number of swollen joints, the body parts corresponding to the number of tender joints or swollen joints, and information related to orthopedic surgery.

[0056] In Modification 1, an example was shown in which a learning model 14M is generated using training data that further includes at least one of the number of tender joints and the number of swollen joints, the sites corresponding to the number of tender joints or swollen joints, and information on orthopedic surgery, but the method is not limited to this. A learning model 14M may also be generated using training data that further includes at least one of the number of tender joints and the number of swollen joints, the sites corresponding to the number of tender joints or swollen joints, and at least one of information on orthopedic surgery.

[0057] Figure 9 is a flowchart illustrating an example of the procedure for generating a learning model. The control unit 11 reads a large amount of training data from the training DB 141, which is linked to the administration information of molecularly targeted drugs, the type and dosage of disease-modifying rheumatic drugs, patient information, and oral glucocorticoid administration information, along with the correct value labels (effect information) (step S101).

[0058] The control unit 11 uses the numerous training data read out to generate a learning model 14M that takes as input information the administration of molecularly targeted drugs, the type and dosage of disease-modifying rheumatic drugs, patient information, and oral glucocorticoid administration information, and outputs effect information (step S102). The control unit 11 stores the generated learning model 14M in the large-capacity storage unit 14 (step S103).

[0059] Figure 10 is a flowchart showing an example of the processing procedure when outputting effect information. When a healthcare professional executes the program of this embodiment, the control unit 21 displays a screen on the display unit 25 for receiving input of molecular targeted drug administration information, disease-modifying rheumatic drug type and dosage, patient information, and oral glucocorticoid administration information (step S201). The control unit 21 accepts input or selection of molecular targeted drug administration information, disease-modifying rheumatic drug type and dosage, patient information, and oral glucocorticoid administration information via input fields, slide switches, pull-down lists, or checkboxes (step S202).

[0060] The control unit 21 determines whether or not it has received a selection of the transmit button (step S203). If the control unit 21 determines that it has not received a selection of the transmit button (step S203: NO), it returns to step S202. If the control unit 21 determines that it has received a selection of the transmit button (step S203: YES), it transmits the administration information of the molecularly targeted drug that was input or selected, the type and dosage of the disease-modifying rheumatoid drug, patient information, and oral glucocorticoid administration information to the server 10 via the communication unit 23 (step S204).

[0061] The communication unit 13 receives information transmitted from the computer 20 regarding the administration of molecularly targeted drugs, the type and dosage of disease-modifying rheumatoid arthritis drugs, patient information, and oral glucocorticoid administration (step S301). The control unit 11 inputs the received information regarding the administration of molecularly targeted drugs, the type and dosage of disease-modifying rheumatoid arthritis drugs, patient information, and oral glucocorticoid administration into the learning model 14M to acquire effect information (step S302).

[0062] The control unit 11 stores the acquired effect information in the large-capacity storage unit 14, associating it with the administration information of the molecularly targeted drug, the type and dosage of the disease-modifying rheumatoid drug, patient information, and the administration information of the oral glucocorticoid (step S303). The communication unit 13 transmits the administration information of the molecularly targeted drug, the type and dosage of the disease-modifying rheumatoid drug, patient information, and the administration information of the oral glucocorticoid, along with the effect information, to the computer 20 (step S304). The communication unit 13 may also transmit only the acquired effect information to the computer 20.

[0063] The communication unit 23 receives information transmitted from the server 10 regarding the administration of molecularly targeted drugs, the type and dosage of disease-modifying rheumatoid drugs, patient information, and information regarding the administration and effects of oral glucocorticoids (step S205). The control unit 21 displays the received information regarding the administration of molecularly targeted drugs, the type and dosage of disease-modifying rheumatoid drugs, patient information, and information regarding the administration and effects of oral glucocorticoids on the display unit 25 (step S206).

[0064] According to Embodiment 1, the information processing system can output efficacy information based on the administration information of molecularly targeted drugs, the type and dosage of disease-modifying rheumatic drugs, patient information, and the administration information of oral glucocorticoids.

[0065] (Embodiment 2) Embodiment 2 describes an example in which an information processing system acquires multiple modified input information, which is obtained by modifying at least one of the following: information on the administration of molecularly targeted drugs, the type and dosage of disease-modifying antirheumatic drugs, information on the administration of oral glucocorticoids, or patient information, and outputs multiple effect information by inputting each modified input information into a learning model.

[0066] The communication unit 23 receives information transmitted from the server 10 regarding the administration of molecularly targeted drugs, the type and dosage of disease-modifying rheumatic drugs, patient information, and information on the administration and effects of oral glucocorticoids. The control unit 21 displays the received information on the administration of molecularly targeted drugs, the type and dosage of disease-modifying rheumatic drugs, patient information, and information on the administration and effects of oral glucocorticoids on the display unit 25.

[0067] Figure 11 is an explanatory diagram showing an example of a screen. The screen d300 in Figure 11 includes a first area d1 and a second area d2. In the first area d1, similar to d22 in Figure 8, information on the administration of molecularly targeted drugs, the type and dosage of disease-modifying rheumatic drugs, patient information, and information on the administration of oral glucocorticoids input into the learning model 14M is displayed. In the second area d2, from top to bottom, there is an effect information field d21, a request field d31, a confirmation button b2, and an output field d32.

[0068] The effect information field d21 displays the effect information (probability of effective treatment) when the information displayed in the first domain d1 is input into the learning model 14M. The request field d31 accepts request information from healthcare professionals. Request information includes, for example, the administration information of molecularly targeted drugs, the type and dosage of disease-modifying rheumatic drugs, patient information, and oral glucocorticoid administration information displayed in the first domain d1, as well as items that are considered important when proposing future treatment plans (hereinafter referred to as "important items") and the numerical range of those important items. If an important item requires a numerical range, the numerical range of the important item is used; if an important item does not require a numerical range, the numerical range of the important item is not used.

[0069] Specifically, healthcare professionals will input the following as important items: "Type and dosage of molecularly targeted drugs," "Dosage of infliximab," "Type and dosage of disease-modifying rheumatoid arthritis drugs," "Dosage of methotrexate," "Dosage of oral glucocorticoids," or "Information regarding orthopedic surgery." If "Dosage of infliximab," "Dosage of methotrexate," or "Dosage of oral glucocorticoids" is entered as an important item, the numerical range for that important item will be entered as "3-10 mg / kg / dose," "0-16 mg / week," or "Z1-Z10 mg / day (prednisolone equivalent)." If "Type and dosage of molecularly targeted drugs," "Type and dosage of disease-modifying rheumatoid arthritis drugs," or "Information regarding orthopedic surgery" is entered as an important item, nothing will be entered in the numerical range for that important item.

[0070] The control unit 21 may pre-exclude non-changeable information such as the patient's gender, age, height, weight, and duration of illness from the priority items and their numerical ranges. If there are no specific requests from healthcare professionals, the request field d31 may be left blank. In the following explanation, we will describe an example where request information is entered in the request field d31. In the request field d31 of Figure 11, the priority items "Infliximab dosage," "Type and dosage of disease-modifying rheumatoid arthritis drugs," and "Information regarding orthopedic surgery," along with the numerical ranges for the priority items "3 to 10 mg / kg / dose," "- (no input)," and "- (no input)," are displayed.

[0071] The confirmation button b2 is used to send the information contained in the first area d1 and the second area d2 to the server 10. The output field d32 displays the output results of the learning model 14M. In Figure 11, the confirmation button b2 is not selected, so the output field d32 is blank.

[0072] The control unit 21 accepts the input of request information in the second area d2 via an input field, slide switch, pull-down list, or checkbox, etc. The control unit 21 determines whether or not the selection of the confirm button b2 has been accepted. If the control unit 21 determines that the selection of the confirm button b2 has not been accepted, it accepts the input of request information again. If the control unit 21 determines that the selection of the confirm button b2 has been accepted, it transmits the administration information of the molecular targeted drug, the type and dosage of the disease-modifying rheumatoid drug, patient information, the administration information of the oral glucocorticoid, and the request information to the server 10 via the communication unit 23.

[0073] The communication unit 13 receives information transmitted from the computer 20 regarding the administration of molecularly targeted drugs, the type and dosage of disease-modifying rheumatic drugs, patient information, information regarding the administration of oral glucocorticoids, and request information. The control unit 11 acquires a plurality of modified input information, each modified by changing at least one of the received information regarding the administration of molecularly targeted drugs, the type and dosage of disease-modifying rheumatic drugs, patient information, and information regarding the administration of oral glucocorticoids. Specifically, the control unit 11 acquires a plurality of modified input information, each modified by random sampling using the important items included in the request information and the numerical range of those important items, each modified by changing at least one of the information regarding the administration of molecularly targeted drugs, the type and dosage of disease-modifying rheumatic drugs, patient information, and information regarding the administration of oral glucocorticoids.

[0074] If the priority item is "dosage of infliximab" and the numerical range for the priority item is "3 to 10 mg / kg / dose", the control unit 11 acquires multiple change input information, changing the type and dosage of the molecular targeted drug (infliximab: x 1 mg / kg / dose) displayed in the first area d1 of screen d300 to "infliximab: 3 to 10 mg / kg / dose". If the priority item is "type and dosage of disease-modifying rheumatoid arthritis drugs", the control unit 11 acquires multiple change input information, changing the type and dosage of the disease-modifying rheumatoid arthritis drug ("methotrexate: y 1 mg / week") displayed in the first area d1 to another type and dosage of disease-modifying rheumatoid arthritis drug (for example, "methotrexate: 0 to 16 mg / week", "tacrolimus: 0 to 3 mg / day", or "iguratimod: 25 mg / day", etc.).

[0075] If the priority item is "information related to orthopedic surgery," the control unit 11 acquires multiple change input information by changing the information related to orthopedic surgery (artificial joint replacement) displayed in the first area d1 to information related to other orthopedic surgery (for example, arthroplasty or arthrodesis). If the priority item is "dosage of oral glucocorticoids" and the numerical range of the priority item is "Z1 to Z10 mg / day," the control unit 11 acquires multiple change input information by changing the oral glucocorticoid dosage (5 mg / day) displayed in the first area d1 to "oral glucocorticoids: Z1 to Z10 mg / day."

[0076] If multiple sets of priority items and numerical ranges for those priority items are entered, the control unit 11 acquires multiple modified input information that changes at least one of the following: the administration information of the corresponding molecular targeted drug, the type and dosage of the disease-modifying rheumatoid drug, patient information, and the administration information of the oral glucocorticoid. If the control unit 11 has not received request information from the computer 20, it acquires multiple modified input information that changes at least one of the following: the administration information of the molecular targeted drug, the type and dosage of the disease-modifying rheumatoid drug, patient information, and the administration information of the oral glucocorticoid, based on the pre-set priority items and numerical ranges for those priority items. In this case, the pre-set priority items and numerical ranges for those priority items do not need to include information that cannot be changed, such as the patient's gender, age, height, weight, and duration of illness.

[0077] The control unit 11 acquires effect information by inputting each acquired change input information into the learning model 14M. The control unit 11 determines whether or not all change input information has been input into the learning model 14M. If the control unit 11 determines that not all change input information has been input into the learning model 14M, it acquires effect information by inputting each acquired change input information into the learning model 14M. If the control unit 11 determines that all change input information has been input into the learning model 14M, it stores the multiple effect information and the multiple change input information in the large-capacity storage unit 14 in association with each other.

[0078] The control unit 11 reads from the large-capacity storage unit 14 the change input information whose effect information is equal to or greater than a predetermined value (for example, 0.8) and the effect information corresponding to that change input information, from among the multiple effect information acquired. The communication unit 13 transmits the read change input information and the effect information corresponding to that change input information to the computer 20.

[0079] The communication unit 23 receives change input information transmitted from the server 10 and effect information corresponding to said change input information. The control unit 21 displays the received change input information and the effect information corresponding to said change input information on the display unit 25.

[0080] Figure 12 is an explanatory diagram showing an example of a screen. The output field d32 of screen d300 displays the change input information for which the effect information is greater than or equal to a predetermined value, and the effect information corresponding to that change input information. In the output field d32 of Figure 12, the changed information (dosage of infliximab, type and dosage of disease-modifying rheumatoid arthritis drugs, and information on orthopedic surgery) is displayed as the important item and the numerical range of that important item among the change input information for which the effect information is greater than or equal to a predetermined value.

[0081] When the control unit 21 receives request information input again, it determines whether the OK button b2 has been selected. If the control unit 21 determines that the OK button b2 has not been selected, it terminates the process. If the control unit 21 determines that the OK button b2 has been selected, it sends the request information that was received again, the molecular targeted drug administration information displayed in the first area d1, the type and dosage of the disease-modifying rheumatoid drug, patient information, oral glucocorticoid administration information, and the request information to the server 10 via the communication unit 23.

[0082] In Embodiment 2, the control unit 11 reads from the large-capacity storage unit 14 the change input information with an effect value of a predetermined amount or more, and the effect information corresponding to that change input information, and transmits the read change input information and the effect information corresponding to that change input information to the computer 20. However, the control unit 11 may transmit the acquired multiple effect information and multiple change input information to the computer 20 via the communication unit 13, regardless of the value of the effect information.

[0083] The following describes the processing in that case. The communication unit 23 receives multiple effect information and multiple change input information transmitted from the server 10. The control unit 21 displays the multiple effect information and multiple change input information on the display unit 25. The output field d32 of the screen d300 displays the multiple change input information and the effect information corresponding to that change input information.

[0084] Figure 13 is a flowchart showing the processing procedure for outputting effect information based on change input information. The flowchart in Figure 13 is the same as the process shown in Figure 10, but with steps S205 onwards replaced by steps S401 to S407 and steps S501 to S507. Steps similar to those in Figure 10 are omitted from explanation.

[0085] The control unit 21 displays the received molecular targeted drug administration information, the type and dosage of disease-modifying rheumatoid drugs, patient information, and oral glucocorticoid administration information and effect information on the display unit 25 (step S401). In the second area d2, the control unit 21 accepts the input of request information via an input field, slide switch, pull-down list, or checkbox, etc. (step S402).

[0086] The control unit 21 determines whether or not it has received a selection of the OK button (step S403). If the control unit 21 determines that it has not received a selection of the OK button (step S403: NO), it returns to step S402. If the control unit 21 determines that it has received a selection of the OK button (step S403: YES), it transmits the administration information of the molecularly targeted drug that was input or selected, the type and dosage of the disease-modifying rheumatoid drug, patient information, administration information of oral glucocorticoids, and request information to the server 10 via the communication unit 23 (step S404).

[0087] The communication unit 13 receives information transmitted from the computer 20 regarding the administration of molecularly targeted drugs, the type and dosage of disease-modifying rheumatic drugs, patient information, information regarding the administration of oral glucocorticoids, and request information (step S501). The control unit 11 obtains a plurality of modified input information by random sampling using important items and numerical ranges of said important items, changing at least one of the information regarding the administration of molecularly targeted drugs, the type and dosage of disease-modifying rheumatic drugs, patient information, and information regarding the administration of oral glucocorticoids (step S502).

[0088] The control unit 11 acquires effect information by inputting each acquired change input information into the learning model 14M (step 503). The control unit 11 determines whether or not all change input information has been input into the learning model 14M (step 504). If the control unit 11 determines that not all change input information has been input into the learning model 14M (step 504: NO), it returns to step S503. If the control unit 11 determines that all change input information has been input into the learning model 14M (step 504: YES), it stores the multiple effect information and the multiple change input information in the large-capacity storage unit 14 in association with each other (step 505).

[0089] The control unit 11 reads from the large-capacity storage unit 14 the change input information whose effect information is equal to or greater than a predetermined value, and the effect information corresponding to that change input information, from among the multiple effect information acquired (step 506). The communication unit 13 transmits the read change input information and the effect information corresponding to that change input information to the computer 20 (step 507).

[0090] The communication unit 23 receives the change input information transmitted from the server 10 and the effect information corresponding to the change input information (step 405). The control unit 21 displays the received change input information and the effect information corresponding to the change input information on the display unit 25 (step 406). When the control unit 21 receives request information input again, it determines whether the OK button has been selected or not (step 407). If the control unit 21 determines that the OK button has not been selected (step 407: NO), it terminates the process. If the control unit 21 determines that the OK button has been selected (step 407: YES), it returns the process to step S404.

[0091] (Modification 2) Embodiment 2 shows an example in which the effect information is displayed on the display unit 25 along with a predetermined amount of modified input information and the effect information corresponding to said modified input information, but it is not limited to this. Input information received on the screen d100 (see Figures 6 and 7) (instruction information of molecularly targeted drugs, type and dosage of disease-modifying rheumatic drugs, administration information of oral glucocorticoids, and patient information) and the effect information corresponding to said input information, as well as modified input information and the effect information corresponding to said modified input information, may be displayed on the display unit 25.

[0092] The following describes the processing in that case. The communication unit 13 receives information transmitted from the computer 20, including information on the administration of molecularly targeted drugs, the type and dosage of disease-modifying rheumatic drugs, patient information, information on the administration of oral glucocorticoids, and request information. The control unit 11 inputs the input information, including the information on the administration of molecularly targeted drugs, the type and dosage of disease-modifying rheumatic drugs, patient information, and information on the administration of oral glucocorticoids, into the learning model 14M to obtain effect information.

[0093] The control unit 11 acquires multiple modified input information by random sampling, changing at least one of the following: information on the administration of molecularly targeted drugs, the type and dosage of disease-modifying rheumatic drugs, patient information, and information on the administration of oral glucocorticoids. The control unit 11 acquires multiple effect information by repeatedly inputting each acquired modified input information into the learning model 14M. The control unit 11 stores the multiple effect information and the multiple modified input information in the large-capacity storage unit 14 in association with each other.

[0094] The control unit 11 reads from the large-capacity storage unit 14 the change input information whose effect information is equal to or greater than a predetermined value, and the effect information corresponding to that change input information, from among the multiple effect information acquired. The communication unit 13 transmits to the computer 20 the input information received on the screen d100 and the effect information corresponding to that input information, as well as the read change input information and the effect information corresponding to that change input information.

[0095] The communication unit 23 receives input information and corresponding effect information transmitted from the server 10, as well as modified input information and corresponding effect information. The control unit 21 displays the received input information and corresponding effect information, as well as modified input information and corresponding effect information, on the display unit 25.

[0096] According to Embodiment 2, the information processing system can obtain multiple modified input information by changing at least one of the following: information on the administration of molecularly targeted drugs, the type and dosage of disease-modifying antirheumatic drugs, the patient information, or the information on the administration of oral glucocorticoids, and by inputting each modified input information into a learning model, it can output multiple effect information.

[0097] According to Embodiment 2, the information processing system can identify change input information whose effect information is equal to or greater than a predetermined value, and output the identified change input information and the effect information corresponding to that change input information.

[0098] According to Embodiment 2, the information processing system can identify among the acquired change input information that has a higher probability of being effective in providing treatment. For example, if the effectiveness information (probability of effective treatment) for "etanercept," "methotrexate: 6 mg / week," "23-year-old female, 164 cm tall, 58 kg in weight," and "oral glucocorticoid: 5 mg / day" is low, the information processing system can identify [golimumab + methotrexate 12.4 mg / week] and [filgotinibuma + methotrexate 8.2 mg / week] as effective treatment methods (change input information).

[0099] (Embodiment 3) Embodiment 3 describes a method by which an information processing system retrains a learning model based on change input information and patient effect information in accordance with said change input information.

[0100] The control unit 21 acquires change input information and patient effectiveness information (hereinafter referred to as "patient effectiveness information") based on the change input information, either through manual input by a medical professional or self-reported by the patient. Patient effectiveness information is, for example, the therapeutic effect when a treatment menu with a high probability of being effective is administered to the patient based on the change input information, and is indicated as either "0 (ineffective)" or "1 (effective)".

[0101] If the above treatment menu is actually implemented for a patient and the symptoms go into remission and the symptom relief lasts for more than six months, the patient's effectiveness information is judged as "1 (effective)". Otherwise, the patient's effectiveness information is judged as "0 (ineffective)". The communication unit 23 transmits the change input information and the patient's effectiveness information to the server 10.

[0102] The communication unit 23 receives the change input information and patient effect information transmitted from the computer 20. The control unit 21 stores the received change input information and patient effect information in the large-capacity storage unit 14. The control unit 21 determines whether a predetermined period (for example, 6 months) has elapsed since the learning model 14M was last relearned.

[0103] If the control unit 21 determines that a predetermined period has not elapsed since the last relearning of the learning model 14M, it stores the change input information received from the computer 20 and the patient's effect information in the large-capacity storage unit 14 until the predetermined period has elapsed. If the control unit 21 determines that a predetermined period has elapsed since the last relearning of the learning model 14M, it reads out multiple sets of change input information and patient effect information combinations from the large-capacity storage unit 14. The control unit 21 relearns the learning model 14M using the read-out sets of change input information and patient effect information combinations. The control unit 21 stores the relearned learning model 14M in the large-capacity storage unit 14.

[0104] In Embodiment 3, an example was shown in which the control unit 11 retrains the learning model 14M based on the period since the learning model 14M was last retrained, but the system is not limited to this. The control unit 11 may also retrain the learning model based on the number of times modified input information, etc., has been stored since the learning model 14M was last retrained.

[0105] The following describes the process in that case. The control unit 11 stores the received change input information and patient effect information in the large-capacity storage unit 14. After the learning model 14M is last retrained, the control unit 11 determines whether the number of times the change input information, etc. has been stored in the large-capacity storage unit 14 is greater than or equal to a predetermined number (for example, 100 times). If the control unit 11 determines that the number of times the change input information, etc. has been stored in the large-capacity storage unit 14 is less than or equal to the predetermined number, it stores the change input information, etc. received from the computer 20 in the large-capacity storage unit 14 until the number of times the change input information, etc. has been stored reaches the predetermined number.

[0106] If the control unit 11 determines that the number of times change input information, etc., has been stored in the large-capacity storage unit 14 is greater than or equal to a predetermined number, it reads out multiple sets of change input information and patient effect information combinations from the large-capacity storage unit 14. The control unit 11 retrains the learning model 14M using the multiple sets of change input information and patient effect information combinations. The control unit 11 stores the retrained learning model 14M in the large-capacity storage unit 14.

[0107] Furthermore, the control unit 11 may retrain the learning model 14M at a predetermined time (for example, at the end of the fiscal year) using multiple combinations of change input information and patient effect information stored in the large-capacity storage unit 14.

[0108] Figure 14 is a flowchart showing an example of a processing procedure related to relearning. The control unit 21 acquires change input information and patient effect information based on manual input by medical professionals or self-reporting by patients (step S601). The communication unit 23 transmits the change input information and patient effect information to the server 10 (step S602).

[0109] The communication unit 23 receives the change input information and patient effect information transmitted from the computer 20 (step S701). The control unit 21 stores the received change input information and patient effect information in the large-capacity storage unit 14 (step S702). The control unit 21 determines whether a predetermined period has elapsed since the learning model 14M was last relearned (step S703). If the control unit 21 determines that a predetermined period has not elapsed since the learning model 14M was last relearned (step S703: NO), it returns to step S601.

[0110] If the control unit 21 determines that a predetermined period has elapsed since the learning model 14M was last relearned (step S703: YES), it reads out multiple sets of change input information and patient effect information combinations from the large-capacity storage unit 14 (step S704). The control unit 21 relearns the learning model 14M using the read-out sets of change input information and patient effect information combinations (step S705). The control unit 21 stores the relearned learning model 14M in the large-capacity storage unit 14 (step S706).

[0111] According to Embodiment 3, the information processing system can retrain its learning model by using a combination of multiple sets of change input information and patient effect information corresponding to said change input information.

[0112] According to Embodiment 3, the information processing system can update the learning model by retraining the learning model using the procedure described above.

[0113] (Embodiment 4) Embodiment 4 describes a method for obtaining modified input information by changing an input item whose contribution to the effect information satisfies predetermined conditions, and for outputting effect information based on the obtained modified input information.

[0114] The communication unit 13 receives information transmitted from the computer 20 regarding the administration of molecularly targeted drugs, the type and dosage of disease-modifying rheumatoid arthritis drugs, patient information, and oral glucocorticoid administration information. The control unit 11 inputs the received information, including the administration of molecularly targeted drugs, the type and dosage of disease-modifying rheumatoid arthritis drugs, patient information, and oral glucocorticoid administration information, into the learning model 14M to acquire effect information.

[0115] The control unit 11 calculates the contribution of each input item (for example, the type and dosage of molecular targeted drugs, the type and dosage of disease-modifying rheumatoid drugs, age, information on orthopedic surgery, and the dosage of oral glucocorticoids, etc.) included in the input information input to the learning model 14M to the output effect information. "Type and dosage of molecular targeted drugs" are, for example, "etanercept: 25 mg / week", "tocilizumab: 162 mg / week", or "infliximab: 3 mg / kg / dose". "Type and dosage of disease-modifying rheumatoid drugs" are, for example, "methotrexate: 6 mg / week", "tacrolimus: 1.5 mg / day", or "iguratimod: 25 mg / day".

[0116] The contribution level is a positive or negative value indicating the degree to which each input item included in the input information contributes positively or negatively to the output of the effect information. It is calculated, for example, by SHAP (SHapley Additive exPlanation), LIME (Local Interpretable Model-Agnostic Explanations), or an Attention mechanism. A larger contribution level for each input item indicates that the input item has a positive effect on the output of the effect information, while a smaller contribution level indicates that the input item has a negative effect. Furthermore, an input item's contribution level approaching zero indicates that it has neither a positive nor a negative effect on the output of the effect information.

[0117] The control unit 11 uses the calculated contribution of each input item to identify input items from the input information input to the learning model 14M whose contribution to the output of effect information by the learning model 14M satisfies predetermined conditions. Input items whose contribution to the output of effect information satisfies predetermined conditions include, for example, the input item with the largest contribution to the output of effect information, the top predetermined number (e.g., 3) of input items with the largest contribution to the output of effect information, input items whose contribution to the output of effect information is equal to or greater than a predetermined value, the input item with the smallest contribution to the output of effect information, the top predetermined number of input items with the smallest contribution to the output of effect information, or input items whose contribution to the output of effect information is equal to or less than a predetermined value. Hereafter, the input item with the smallest contribution to the output of effect information will be used as an example. The control unit 11 may also exclude input items that cannot be changed, such as the patient's gender, age, height, weight, and duration of illness, before identifying input items whose contribution to the output of effect information by the learning model 14M satisfies predetermined conditions.

[0118] The control unit 11 acquires change input information corresponding to the identified input item. Specifically, if the identified input item is "Infliximab: x 1 mg / kg / dose (type and dosage of molecular targeted drug)", the control unit 11 acquires change input information that changes "Infliximab: x 1 mg / kg / dose" to "Infliximab: 3-10 mg / kg / dose", "Etanercept: 25 mg / week (type and dosage of other molecular targeted drugs)", or "Do not use", etc. If the identified input item is "Methotrexate: y 1 mg / week (type and dosage of disease-modifying rheumatoid arthritis drug)", the control unit 11 acquires change input information that changes "Methotrexate: y 1 mg / week" to "Methotrexate: 0-16 mg / week", "Tacrolimus: 0-3 mg / day (type and dosage of other disease-modifying rheumatoid arthritis drugs)", or "Do not use", etc.

[0119] If the identified input item is "arthrodesis (information related to orthopedic surgery)", the control unit 11 acquires modified input information by changing "arthroplasty" to "arthroplasty", "arthrodesis", "synovectomy", "spinal surgery", or "no surgery", etc. If the identified input item is "oral glucocorticoid: z1 mg / day (dosage of oral glucocorticoid)", the control unit 11 acquires modified input information by changing "oral glucocorticoid: z1 mg / week" to "oral glucocorticoid: Z1 to Z10 mg / week (dosage of other oral glucocorticoids)", or "not used", etc. Furthermore, if there are multiple identified input items, the control unit 11 acquires modified input information by changing multiple input items.

[0120] The control unit 11 inputs the acquired change input information to the learning model 14M, thereby acquiring the effect information output by the learning model 14M. The communication unit 13 transmits the acquired effect information and the change input information corresponding to the effect information to the computer 20.

[0121] The communication unit 23 receives effect information and corresponding change input information transmitted from the server 10. The control unit 21 displays the received effect information and corresponding change input information on the display unit 25. For example, the control unit 21 displays the screen shown in Figure 8 on the display unit 25. In this case, the effect information field d21 displays the effect information output by the learning model 14M. The input data field d22 displays the change input information corresponding to the effect information output in the effect information field d21.

[0122] If there are multiple effect information outputs from the learning model 14M, the multiple effect information will be displayed in the effect information field d21, and the multiple change input information will be displayed in the input data field d22. In addition, among the input items of the change input information displayed in the input data field d22, the changed input item (the input item with the smallest contribution to the output of the effect information) may be indicated by hatching or bold text. The control unit 21 may display only the effect information transmitted from the server 10 in the display unit 25.

[0123] Figure 15 is a flowchart showing the processing procedure for outputting effect information based on change input information. The flowchart in Figure 15 is the same as the process shown in Figure 10, but with steps S302 onwards replaced by steps S801 to S805 and steps S901 to S902. Steps similar to those in Figure 10 are omitted from explanation.

[0124] The control unit 11 calculates the contribution of each input item (e.g., "etanercept: 25 mg / week" and "methotrexate: 16 mg / week") included in the input information input to the learning model 14M (e.g., information on the administration of molecularly targeted drugs) to the output effect information (step S801). Based on the calculated contribution of each input item, the control unit 11 identifies the input item among the input items of the input information input to the learning model 14M that has the smallest contribution to the output of effect information by the learning model 14M (step S802).

[0125] The control unit 11 acquires change input information corresponding to the identified input item (step S803). The control unit 11 inputs the acquired change input information to the learning model 14M and acquires the effect information output by the learning model 14M (step S804). The communication unit 13 transmits the acquired effect information and the change input information corresponding to the effect information to the computer 20 (step S805).

[0126] The communication unit 23 receives effect information and corresponding change input information transmitted from the server 10 (step S901). The control unit 21 displays the received effect information and corresponding change input information on the display unit 25 (step S902).

[0127] According to Embodiment 4, the information processing system can identify input items from among the input information, including information on the administration of molecularly targeted drugs, the type and dosage of disease-modifying antirheumatic drugs, patient information, and information on the administration of oral glucocorticoids, that satisfy predetermined conditions for their contribution to the output of effect information by the learning model.

[0128] According to Embodiment 4, the information processing system can obtain change input information corresponding to the identified input item and output effect information by inputting the obtained change input information into the learning model.

[0129] The matters described in each of the embodiments described above can be combined with each other. Furthermore, the independent claims and dependent claims described in the claims can be combined with each other in any combination, regardless of the form of reference. In addition, although the claims use a form in which claims referencing two or more other claims (multi-claim form), they are not limited to this. They may also be described using a form in which multi-claims referencing at least one multi-claim (multi-multi-claim).

[0130] The embodiments disclosed herein should be considered in all respects to be illustrative and not restrictive. The scope of the invention is indicated by the claims and not in the sense described above, and all modifications within the sense and scope equivalent to the claims are intended to be included.

[0131] 10 Information Processing Device (Server) 11 Control Unit 12 Storage Unit 12P Control Program 13 Communication Unit 14 Large Capacity Storage Unit 14M Learning Model 141 Training DB 15 Reading Unit 20 Information Processing Device (Computer) 21 Control Unit 22 Storage Unit 22P Control Program 23 Communication Unit 24 Input Unit 25 Display Unit 1a Portable Storage Medium d100 Screen b1 Send Button d200 Screen d21 Effect Information Field d22 Input Data Field d300 Screen d1 First Area d2 Second Area b2 Confirm Button d31 Request Field d32 Output Field

Claims

1. A program that causes a computer to execute a process to output effect information when the acquired molecular targeted drug administration information, disease-modifying antirheumatic drug type and dosage, patient information, and oral glucocorticoid administration information is input into a learning model that outputs effect information regarding the therapeutic effect of rheumatoid arthritis when the molecular targeted drug administration information, disease-modifying antirheumatic drug type and dosage, patient information, and oral glucocorticoid administration information is input into the learning model.

2. The program according to claim 1, wherein the molecular targeted drug comprises at least one of a biological agent and a JAK inhibitor, and the administration information of the molecular targeted drug includes whether or not the molecular targeted drug was administered, the type of molecular targeted drug, the dosage and the date of administration commencement, and the date and reason for discontinuation of the molecular targeted drug.

3. The program according to claim 1, wherein the disease-modifying antirheumatic drug includes methotrexate and disease-modifying antirheumatic drugs other than methotrexate.

4. The program according to claim 1, wherein the patient information includes age, sex, height, weight, duration of illness, antibody titer of anti-CCP antibody, antibody titer of rheumatoid factor, blood test data, physical disability assessment index for rheumatoid arthritis, and disease activity assessment index for rheumatoid arthritis.

5. The program according to claim 1, wherein the information on the administration of oral glucocorticoids includes the dosage and start date of administration of oral glucocorticoids, as well as the date and time of discontinuation of oral glucocorticoids and the reason for discontinuation.

6. The program according to claim 4, further comprising, the patient information, at least one of the number of tender joints and the number of swollen joints, the site corresponding to the number of tender joints or the number of swollen joints, and information relating to orthopedic surgery.

7. The program according to claim 6, which obtains a plurality of modified input information by changing at least one of the following: information on the administration of a molecularly targeted drug, the type and dosage of a disease-modifying antirheumatic drug, the patient information, or the information on the administration of an oral glucocorticoid, and outputs a plurality of effect information by inputting each modified input information into a learning model.

8. The program according to claim 7, which identifies change input information whose effect information is equal to or greater than a predetermined value, and outputs the identified change input information and the effect information corresponding to the change input information.

9. The program according to claim 7, which acquires patient effect information in accordance with the change input information, and retrains the learning model based on the change input information and the acquired effect information.

10. The program according to claim 1, which identifies input items from among the input information, including information on the administration of molecularly targeted drugs, the type and dosage of disease-modifying antirheumatic drugs, patient information, and information on the administration of oral glucocorticoids, that satisfy predetermined conditions for the contribution of the learning model to the output of effect information, obtains modified input information corresponding to the identified input items, and outputs effect information by inputting the obtained modified input information into the learning model.

11. A method for generating a learning model in which a computer performs a process to generate a learning model that outputs information on the therapeutic effect of rheumatoid arthritis when information on the administration of molecularly targeted drugs, the type and dosage of disease-modifying antirheumatic drugs, patient information, and oral glucocorticoid administration is input.

12. An information processing method in which a computer performs a process to output effect information by inputting the acquired information on the administration of molecularly targeted drugs, the type and dosage of disease-modifying antirheumatic drugs, patient information, and oral glucocorticoids into a learning model that outputs effect information regarding the therapeutic effect of rheumatoid arthritis when the information on the administration of molecularly targeted drugs, the type and dosage of disease-modifying antirheumatic drugs, patient information, and oral glucocorticoids is input.

13. An information processing device having a control unit, wherein the control unit acquires information on the administration of molecularly targeted drugs, the type and dosage of disease-modifying antirheumatic drugs, patient information of the patient, and information on the administration of oral glucocorticoids, and outputs information on the effect of treatment for rheumatoid arthritis by inputting the acquired information on the administration of molecularly targeted drugs, the type and dosage of disease-modifying antirheumatic drugs, patient information of the patient, and information on the administration of oral glucocorticoids into a learning model that outputs effect information on the treatment effect of rheumatoid arthritis when the information on the administration of molecularly targeted drugs, the type and dosage of disease-modifying antirheumatic drugs, patient information of the patient, and information on the administration of oral glucocorticoids is input into the information processing device.

Images