Information processor, information processing method and computer program

Abstract

To easily and accurately predict, whether a patient is a specific migraine patient for whom prescription of triptan formulation is effective.SOLUTION: An information processor which predicts whether a patient is a specific migraine patient for whom prescription of triptan formulation is effective, from headache patients, comprises: a model acquisition part; an object patient information acquisition part; and a prediction execution part. The model acquisition part acquires a drug effect determination model which is a machine learning model which has: a reply about multiple inquiry items as input; and a determination result of whether a patient is a specific migraine patient, as output. The object patient information acquisition part acquires a reply about multiple inquiry items which are for the object patient. The prediction execution part predicts whether the object patient is the specific migraine patient using, the reply which is for the object patient, and the drug effect determination model, and outputs the prediction result.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The technology disclosed in this specification relates to information processing for predicting whether or not a migraine patient would benefit from the prescription of a triptan drug. [Background technology]

[0002] Migraine is a throbbing headache that occurs on one or both sides of the head. In Japan, the prevalence of migraine is relatively high at 8% (see, for example, Non-Patent Document 1). Triptan preparations are known as drugs that are effective against migraine.

[0003] The International Headache Criteria are used as diagnostic criteria for migraine. Each medical institution interviews headache patients and diagnoses migraine by referring to the criteria. [Prior art documents] [Non-patent literature]

[0004] [Non-Patent Document 1] Hisanori Furukawa and 1 other, "Progression and natural history of migraine", Clinical Neurology, Japanese Society of Neurology, 2011, Vol. 51, No. 11, pp. 1147-1149 Summary of the Invention [Problem to be solved by the invention]

[0005] The above-mentioned International Headache Criteria are not clearly quantifiable standards and are subject to change from time to time. Furthermore, medical institutions often conduct interviews in their own way. Furthermore, there are cases of migraine for which triptan preparations are ineffective. Therefore, there is a problem that it is difficult to accurately determine whether a headache patient is a migraine patient and whether the prescription of a triptan preparation is effective or not.

[0006] This specification discloses a technique that can solve the above-mentioned problems. [Means for solving the problem]

[0007] The technology disclosed in this specification can be realized, for example, in the following forms.

[0008] (1) The information processing device disclosed in this specification is a device that predicts whether a headache patient is a specific migraine patient for whom the prescription of a triptan preparation is effective, and includes a model acquisition unit, a target patient information acquisition unit, and a prediction execution unit. The model acquisition unit acquires a drug efficacy determination model, which is a machine learning model that receives answers to a plurality of questionnaire items as input and outputs a determination result of whether the patient is a specific migraine patient. The target patient information acquisition unit acquires answers to a plurality of questionnaire items targeted at the target patient. The prediction execution unit predicts whether the target patient is a specific migraine patient using the answers targeted at the target patient and the drug efficacy determination model, and outputs the prediction result.

[0009] According to the information processing device, it is possible to predict whether or not a target patient is a specific migraine patient based on the answers to a plurality of interview items of the target patient. Therefore, according to the information processing device, it is possible to easily and accurately predict whether or not a specific migraine patient is one for whom the prescription of a triptan preparation is effective.

[0010] (2) In the information processing device, the plurality of interview items may include an item related to pain exacerbation and amelioration factors. With this configuration, it is possible to predict with even higher accuracy whether a specific migraine patient is one for whom prescription of a triptan preparation is effective.

[0011] (3) In the information processing device, the plurality of interview items may include at least one of an item regarding onset time, an item regarding medical history and / or family history, and an item regarding pain itself. With this configuration, it is possible to predict with even higher accuracy whether a specific migraine patient is one for whom prescription of a triptan preparation is effective.

[0012] (4) In the information processing device, the weighting coefficient of the item related to the pain exacerbation / relief factor among the plurality of interview items in the drug efficacy assessment model may be greater than the weighting coefficients of the other items. With this configuration, it is possible to predict with even higher accuracy whether or not a prescription of a triptan preparation is effective for a specific migraine patient.

[0013] (5) In the information processing device, the model acquisition unit may be configured to acquire a reading model, which is a machine learning model that receives an image of a medical questionnaire including the plurality of medical questionnaire items as an input and outputs answers to the plurality of medical questionnaire items, and the target patient information acquisition unit may be configured to acquire answers to the plurality of medical questionnaire items for the target patient using the image of the medical questionnaire of the target patient and the reading model. With this configuration, answers to the plurality of medical questionnaire items by the target patient can be easily and accurately acquired.

[0014] The technology disclosed in this specification can be realized in various forms, such as an information processing device, an information processing method, a computer program for realizing those methods, a non-transitory recording medium on which that computer program is recorded, etc. [Brief description of the drawings]

[0015] [Figure 1] FIG. 1 is an explanatory diagram conceptually illustrating a headache diagnosis model MO according to the present embodiment. [Diagram 2] FIG. 1 is an explanatory diagram showing a schematic configuration of an information processing device. [Diagram 3] A flowchart showing a reading model acquisition process according to the present embodiment. [Figure 4] An explanatory diagram showing an example of a medical interview item included in the medical interview sheet IS. [Diagram 5] A flowchart showing a drug efficacy assessment model acquisition process according to the present embodiment. [Figure 6] FIG. 13 is an explanatory diagram showing an example of each interview item and its weighting coefficient. [Figure 7] 1 is a flowchart showing a prediction process according to the present embodiment. [Figure 8] An explanatory diagram showing the results of the image reading process of the medical questionnaire IS using the reading model MO1. [Figure 9] An explanatory diagram showing the results of the image reading process of the medical questionnaire IS using the reading model MO1. [Figure 10] An explanatory diagram showing the results of predicting whether a patient is a specific migraine patient using the drug efficacy assessment model MO2. [Figure 11] An explanatory diagram showing the results of predicting whether a patient is a specific migraine patient using the drug efficacy assessment model MO2. [Figure 12] An explanatory diagram showing the relationship between the weighting coefficient of each interview item and prediction accuracy. [Figure 13] An explanatory diagram showing the relationship between the weighting coefficient of each interview item and prediction accuracy. [Figure 14] An explanatory diagram showing the relationship between the weighting coefficient of each interview item and prediction accuracy. [Figure 15] An explanatory diagram showing the relationship between the weighting coefficient of each interview item and prediction accuracy. [Figure 16] An explanatory diagram showing the relationship between the weighting coefficient of each interview item and prediction accuracy. [Figure 17] An explanatory diagram showing the relationship between the weighting coefficient of each interview item and prediction accuracy. [Figure 18] An explanatory diagram showing the relationship between the weighting coefficient of each interview item and prediction accuracy. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0016] A. Embodiment: A-1. Overview of the headache diagnostic model MO: First, an overview of the headache diagnostic model MO in this embodiment will be described. Fig. 1 is an explanatory diagram conceptually showing the headache diagnostic model MO in this embodiment. The headache diagnostic model MO is a model for diagnosing (or assisting in diagnosis, the same applies below) headache patients. The headache diagnostic model MO includes a reading model MO1 and a drug efficacy determination model MO2.

[0017] The reading model MO1 is a model for reading a patient's answers to multiple medical questionnaire items included in the medical questionnaire IS from the medical questionnaire IS filled out by the patient. The reading model MO1 is a machine learning model that receives an image of the medical questionnaire IS including multiple medical questionnaire items as input and outputs the patient's answers to the multiple medical questionnaire items (the reading result RR shown in FIG. 1). Note that in this specification, machine learning is a general term for techniques and methods for finding rules and patterns by using a computer to learn based on a large amount of data (i.e., in a data-driven manner), and includes deep learning.

[0018] The efficacy determination model MO2 is a model for predicting whether a patient is a migraine patient for whom the prescription of triptan preparation TP is effective (hereinafter referred to as a "specific migraine patient"). A specific migraine patient is, for example, a migraine patient for whom a doctor's prescription of triptan preparation TP is effective. The efficacy determination model MO2 is a machine learning model that receives as input answers to a plurality of questionnaire items included in the questionnaire IS (in this embodiment, the read result RR shown in FIG. 1) and outputs a determination result as to whether the patient is a specific migraine patient. The answers to the questionnaire IS are provided, for example, by the patient himself / herself, the patient's representative, a medical professional, etc. The efficacy determination model MO2 may also be a model that identifies headache type in addition to predicting specific migraine patients.

[0019] By using the headache diagnostic model MO of this embodiment, it is possible to accurately predict, based on the questionnaire IS filled out by the patient, whether or not the patient is a specific migraine patient for whom the prescription of the triptan preparation TP would be effective.

[0020] It should be noted that diagnosis using the headache diagnostic model MO of this embodiment can be performed at a facility other than a medical institution, for example, by the patient himself / herself.

[0021] A-2. Configuration of information processing device 100: Next, a description will be given of the configuration of an information processing device 100 for creating a headache diagnostic model MO and for executing prediction processing using the headache diagnostic model MO. Figure 2 is an explanatory diagram showing a schematic configuration of the information processing device 100. The information processing device 100 is formed, for example, by a computer (PC, server, etc.).

[0022] The information processing device 100 includes a control unit 110, a storage unit 120, a display unit 130, an operation input unit 140, and an interface unit 150. These units are connected to each other so as to be able to communicate with each other via a bus 190. The information processing device 100 may include a speaker as an output unit.

[0023] The display unit 130 of the information processing device 100 is, for example, a liquid crystal display, and displays various images and information. The operation input unit 140 is, for example, a keyboard, a mouse, buttons, a microphone, a track pad, and the like, and receives operations and instructions from an administrator. The display unit 130 may be provided with a touch panel to function as the operation input unit 140. The interface unit 150 is, for example, a LAN interface, a USB interface, and the like, and communicates with other devices via wired or wireless communication.

[0024] The storage unit 120 of the information processing device 100 is composed of, for example, a ROM, a RAM, a hard disk drive (HDD), etc., and is used to store various programs and data, and as a work area when executing various programs and a temporary storage area for data. For example, the storage unit 120 stores a prediction processing program CP, which is a computer program for executing various processes described later. The prediction processing program CP is provided in a state stored in a computer-readable recording medium (not shown), such as a CD-ROM, a DVD-ROM, or a USB memory, or is provided in a state in which it can be acquired from an external device (a server or other terminal device on a network) via the interface unit 150, and is stored in the storage unit 120 in a state in which it can operate on the information processing device 100.

[0025] Furthermore, learning data LD, a headache diagnostic model MO, and prediction result data RD are stored in advance or during the execution of various processes described below in the storage unit 120 of the information processing device 100. The contents of this information and data will be explained in conjunction with the explanation of the various processes described below.

[0026] The control unit 110 of the information processing device 100 is configured with, for example, a CPU, and controls the operation of the information processing device 100 by executing a computer program read from the storage unit 120. For example, the control unit 110 functions as a prediction processing unit 111 for executing various processes described below by reading a prediction processing program CP from the storage unit 120 and executing it. The prediction processing unit 111 includes an original data acquisition unit 112, a learning data acquisition unit 113, a model acquisition unit 114, a target patient information acquisition unit 115, and a prediction execution unit 116. The functions of each of these units will be described in conjunction with the description of various processes described below.

[0027] A-3. Read model acquisition process: Next, the reading model acquisition process executed by the information processing device 100 of this embodiment will be described. Fig. 3 is a flowchart showing the reading model acquisition process in this embodiment. The reading model acquisition process is a process for acquiring a reading model MO1, which is a machine learning model for reading a patient's response from a medical questionnaire IS. In this embodiment, the information processing device 100 acquires the reading model MO1 by creating the reading model MO1 by a predetermined machine learning. The reading model acquisition process is started in response to a user inputting a start instruction by operating the operation input unit 140 of the information processing device 100.

[0028] First, the original data acquisition unit 112 (FIG. 2) of the information processing device 100 acquires information (hereinafter referred to as "original data Io1") used to create the reading model MO1 (S110). The original data Io1 is information that is the basis of learning data LD used for training, verifying, and testing the reading model MO1. Specifically, the original data Io1 includes images of medical questionnaires IS filled out by multiple patients. The original data Io1 is acquired via the interface unit 150 or via the operation input unit 140.

[0029] 4 is an explanatory diagram showing an example of medical interview items included in the medical interview sheet IS. As shown in FIG. 4, the medical interview items included in the medical interview sheet IS are classified into, for example, the following four categories. (1) Items related to the time of onset (2) Medical and / or family history (3) Items related to pain itself (4) Items related to factors that increase or reduce pain

[0030] Examples of the above "(1) Items related to the time of onset" include the following items: Onset of the disease occurred between the ages of 10 and 19. (Hereinafter referred to as "Onset_10") Onset of symptoms occurred several weeks ago. (Hereinafter referred to as "Onset_weeksago.") Onset of symptoms occurred several days ago. (Hereinafter referred to as "Onset_daysago.")

[0031] Examples of the above “(2) Items related to medical history and / or family history” include the following: - Have a family member with the same condition. (Hereinafter referred to as "FH_yes") - The mother has the same symptoms. (Hereinafter referred to as "FH_mother.") - History of hypertension (hereinafter referred to as "HT").

[0032] Examples of the above "(3) Items related to pain itself" include the following: The headache is throbbing in nature. (Hereinafter referred to as "Pounding_pain.") The headache has an electrifying quality. (Hereinafter referred to as "Electrifying pain.") - Headaches occur 15 or more times a month. (Hereinafter referred to as "15-_month.")

[0033] Examples of the above "(4) Items related to pain exacerbation and amelioration factors" include the following items. - Headaches worsen with body movement. (Hereinafter referred to as "Body_motion") - Light and sound sensitivity (hereinafter referred to as "Sensitivity_light_sound"). Nausea and vomiting occur during headaches. (Hereafter, this is also referred to as "vomit_nausia.") - Headaches are aggravated by weather. (Hereinafter referred to as "Weather") - Headaches worsening due to menstruation (hereinafter referred to as "Menstral_periods").

[0034] Next, the learning data acquisition unit 113 (FIG. 2) of the information processing device 100 performs preprocessing on the original data Io1 to acquire learning data LD for creating the reading model MO1 (S120). The preprocessing may include, for example, labeling and data expansion. The learning data LD is data in which an image of the medical questionnaire IS filled out by the patient is associated with the patient's answer (for example, "yes" or "no") to each medical question included in the medical questionnaire IS.

[0035] Next, the model acquisition unit 114 (FIG. 2) of the information processing device 100 creates the reading model MO1 by machine learning using the learning data LD for creating the reading model MO1 (S130). Various known machine learning algorithms can be used for the machine learning for creating the reading model MO1. For example, YOLO or the like may be used for creating the reading model MO1.

[0036] The reading model MO1 created by machine learning is stored in the storage unit 120 of the information processing device 100. This completes the process of acquiring the reading model MO1 (FIG. 3).

[0037] A-4. Drug efficacy assessment model acquisition process: Next, a description will be given of a drug efficacy judgment model acquisition process executed by the information processing device 100 of this embodiment. FIG. 5 is a flowchart showing the drug efficacy judgment model acquisition process in this embodiment. The drug efficacy judgment model acquisition process is a process for acquiring a drug efficacy judgment model MO2, which is a machine learning model for determining whether or not a patient is a specific migraine patient for whom the prescription of a triptan preparation TP is effective, based on the patient's answers to a plurality of questionnaire items included in the questionnaire IS. In this embodiment, the information processing device 100 acquires the drug efficacy judgment model MO2 by creating the drug efficacy judgment model MO2 by a predetermined machine learning. The drug efficacy judgment model acquisition process is started in response to a user inputting a start instruction by operating the operation input unit 140 of the information processing device 100.

[0038] First, the original data acquisition unit 112 (FIG. 2) of the information processing device 100 acquires information (hereinafter referred to as "original data Io2") used to create the efficacy judgment model MO2 (S210). The original data Io2 is information that is the basis of learning data LD used for training, verifying, and testing the efficacy judgment model MO2. Specifically, the original data Io2 includes responses by the patient to a plurality of questionnaire items included in the questionnaire IS, and information indicating whether the patient is a specific migraine patient or not. Note that in this embodiment, a patient being a specific migraine patient includes the following two cases. (1) Patients who were diagnosed with migraine at the first visit, took the triptan preparation TP, and self-reported efficacy at the follow-up visit one month later and continued to receive the prescription (2) Patients who were diagnosed with migraine at their first visit, were given a triptan preparation TP, and self-reported that the drug was ineffective at their follow-up visit one month later, but changed their prescription to a different type of triptan preparation TP, self-reported that the drug was effective at their next follow-up visit, and were continued to receive the prescription.

[0039] Next, the learning data acquisition unit 113 (FIG. 2) of the information processing device 100 performs preprocessing on the original data Io2 to acquire learning data LD for creating the drug efficacy assessment model MO2 (S220). As the preprocessing, for example, setting the number of features (medical questionnaire items) is performed. The learning data LD is data in which the patient's answer (for example, "yes" or "no") to each medical questionnaire item included in the medical questionnaire IS is associated with whether the patient is a specific migraine patient or not.

[0040] Next, the model acquisition unit 114 (FIG. 2) of the information processing device 100 creates a drug efficacy judgment model MO2 by machine learning using the learning data LD for creating the drug efficacy judgment model MO2 (S230). For the machine learning for creating the drug efficacy judgment model MO2, various known machine learning algorithms can be used. For example, logistic regression, support vector machine, random forest, LightGBM, etc. may be used for creating the drug efficacy judgment model MO2.

[0041] In this embodiment, each medical question item is weighted. Fig. 6 is an explanatory diagram showing an example of each medical question item and its weighting coefficient. In Fig. 6, each medical question item is arranged from the top in descending order of weighting coefficient.

[0042] The efficacy determination model MO2 created by machine learning is stored in the storage unit 120 of the information processing device 100. This completes the acquisition process (FIG. 5) of the efficacy determination model MO2. When creating the reading model MO1 and the efficacy determination model MO2, for example, a part of the learning data LD may be used as training data for updating model parameters (weights, etc.), another part of the learning data LD may be used as verification data for setting hyperparameters, and another part of the learning data LD may be used as test data for checking the generalization performance of the model.

[0043] A-5. Prediction processing: Next, the prediction process executed by the information processing device 100 of this embodiment will be described. Fig. 7 is a flowchart showing the prediction process in this embodiment. The prediction process is a process for predicting whether or not a target patient is a specific migraine patient for whom the prescription of a triptan formulation TP is effective, using a headache diagnosis model MO including a reading model MO1 and a drug efficacy judgment model MO2. The prediction process is started in response to a user operating the operation input unit 140 of the information processing device 100 to input a start instruction.

[0044] First, the target patient information acquisition unit 115 (FIG. 2) of the information processing device 100 acquires an image of the medical questionnaire IS filled out by the target patient (S310). The image of the medical questionnaire IS is acquired via the interface unit 150 or via the operation input unit 140, and is stored in the storage unit 120. Note that the information processing device 100 may have a scanner function and scan the medical questionnaire IS to acquire the image of the medical questionnaire IS.

[0045] Next, the prediction execution unit 116 (FIG. 2) of the information processing device 100 uses the image of the medical questionnaire IS and the reading model MO1 to obtain answers from the target patient to the multiple medical questionnaire items included in the medical questionnaire IS (S320). That is, the prediction execution unit 116 inputs the image of the medical questionnaire IS to the reading model MO1, and obtains the reading result RR of the medical questionnaire IS output from the reading model MO1 (see FIG. 1).

[0046] Next, the prediction execution unit 116 (FIG. 2) of the information processing device 100 executes a prediction as to whether or not the target patient is a specific migraine patient, using the read result RR of the image of the questionnaire IS and the drug efficacy determination model MO2 (S330). That is, the prediction execution unit 116 inputs the read result RR (the target patient's answers to each question item) into the drug efficacy determination model MO2, thereby obtaining a prediction result as to whether or not the target patient is a specific migraine patient, which is output from the drug efficacy determination model MO2. The prediction execution unit 116 generates prediction result data RD, which is information indicating the prediction result, and stores it in the storage unit 120 of the information processing device 100.

[0047] Next, the prediction execution unit 116 outputs the prediction result (diagnosis result) based on the prediction result data RD (S340). For example, the prediction execution unit 116 causes the display unit 130 to display the prediction result. With the above, the prediction process is completed.

[0048] A-6. Example: An embodiment of the above-mentioned headache diagnostic model MO will be described below.

[0049] 8 and 9 are explanatory diagrams showing the results of the reading process (S320 in FIG. 7) of the image of the medical questionnaire IS using the reading model MO1. As shown in FIG. 8, in this embodiment, the reading process was performed using five data sets (DS1 to DS5) of different types of medical questionnaire IS. The total number of sheets of the medical questionnaire IS was 4,280, and the total number of medical questionnaire items was 205,836. FIG. 9 shows, from the top to the bottom, a confusion matrix showing the relationship between the correct answer value and the predicted value for each of the data sets DS1 to DS5. As shown in FIG. 8 and FIG. 9, in this embodiment, the answers of the patient could be read from the image of the medical questionnaire IS with an extremely high accuracy of 99% or more.

[0050] 10 and 11 are explanatory diagrams showing the prediction results of whether or not a patient is a specific migraine headache patient using the drug efficacy judgment model MO2. In this embodiment, four machine learning algorithms were used: logistic regression (LR), support vector machine (SVM), random forest (RF), and LightGBM (LGBM). FIG. 10 shows the results of cross-validation using data of n=270. FIG. 11 shows the results of validation using data of n=109. Regardless of which machine learning algorithm was used, it was possible to predict whether or not a patient was a specific migraine headache patient with extremely high accuracy, with ROC-AUC of approximately 0.9 or more.

[0051] 12 to 18 are explanatory diagrams showing the relationship between the weighting coefficient of each interview item and the prediction accuracy. In each diagram, each interview item is classified into the above-mentioned four categories, and the weighting coefficient of each interview item is shown, along with the prediction accuracy (ROC-AUC value) when the weighting is adopted. Logistic regression (LR) was used as the machine learning algorithm.

[0052] Figure 12 shows the prediction accuracy when the weighting of each interview item is optimized (hereinafter referred to as the "optimal state"). In this case, the ROC-AUC value was 0.930485.

[0053] Figure 13 shows the prediction accuracy when the weighting coefficients of each interview item are equalized. In this case, the ROC-AUC value was 0.890293, which was approximately 0.04 lower than the optimal state.

[0054] Figure 14 shows the prediction accuracy when the weighting coefficient of the questionnaire item related to onset time is set to zero, compared to the state shown in Figure 13 where the weighting coefficients of each questionnaire item are uniform. In this case, the ROC-AUC value was 0.900910, a decrease of approximately 0.03 from the optimal state.

[0055] Figure 15 shows the prediction accuracy when the weighting coefficients of the questionnaire items related to medical history and / or family history are set to zero, compared to the state shown in Figure 13 where the weighting coefficients of each questionnaire item are uniform. In this case, the ROC-AUC value was 0.878665, a decrease of about 0.06 from the optimal state.

[0056] Figure 16 shows the prediction accuracy when the weighting coefficient of the interview item related to pain itself is set to zero, compared to the state shown in Figure 13 where the weighting coefficients of each interview item are uniform. In this case, the ROC-AUC value was 0.898635, a decrease of approximately 0.03 from the optimal state.

[0057] Figure 17 shows the prediction accuracy when the weighting coefficients of the interview items related to pain exacerbation / relief factors are set to zero, compared to the state shown in Figure 13 where the weighting coefficients of each interview item are uniform. In this case, the ROC-AUC value was 0.728766, a decrease of approximately 0.20 from the optimal state.

[0058] Figure 18 shows the prediction accuracy when the weighting coefficient of the interview item related to pain exacerbation and remission factors is doubled from the state in Figure 13 where the weighting coefficient of each interview item is uniform. In this case, the ROC-AUC value was 0.916077, which was about 0.01 lower than the optimal state.

[0059] 12 to 18, it can be said that the prediction accuracy is relatively greatly reduced when the weighting coefficient of the interview item related to the pain aggravation / remission factor is reduced, and the prediction accuracy is improved when the weighting coefficient of the interview item related to the pain aggravation / remission factor is increased. Therefore, it can be said that the weighting coefficient of the interview item related to the pain aggravation / remission factor is preferably larger than the weighting coefficients of the other interview items.

[0060] A-7. Advantages of this embodiment: As described above, the information processing device 100 of this embodiment is a device for predicting whether or not a patient is a specific migraine patient for whom the prescription of the triptan preparation TP is effective, and includes a model acquisition unit 114, a target patient information acquisition unit 115, and a prediction execution unit 116. The model acquisition unit 114 acquires a drug efficacy determination model MO2, which is a machine learning model that receives answers by a patient to a plurality of interview items and outputs a determination result of whether or not the patient is a specific migraine patient. The target patient information acquisition unit 115 acquires answers by the target patient to a plurality of interview items. The prediction execution unit 116 predicts whether or not the target patient is a specific migraine patient using the answers by the target patient and the drug efficacy determination model MO2, and outputs the prediction result.

[0061] In this way, the information processing device 100 of this embodiment can predict whether or not the target patient is a specific migraine patient based on the target patient's answers to a plurality of interview items. Therefore, the information processing device 100 of this embodiment can easily and accurately predict whether or not the target patient is a specific migraine patient for whom the prescription of the triptan preparation TP is effective.

[0062] In addition, in the information processing device 100 of this embodiment, the multiple interview items may include items related to pain exacerbation and remission factors. By adopting such a configuration, it is possible to predict with even higher accuracy whether a specific migraine patient is one for whom the prescription of the triptan preparation TP is effective.

[0063] In addition, in the information processing device 100 of the present embodiment, the multiple interview items may include at least one of an item regarding onset time, an item regarding medical history and / or family history, and an item regarding pain itself. By adopting such a configuration, it is possible to predict with even higher accuracy whether or not a specific migraine patient is a patient for whom the prescription of the triptan preparation TP is effective.

[0064] In the information processing device 100 of this embodiment, the weighting coefficient of the item related to the pain exacerbation / relief factor among the multiple interview items in the drug efficacy judgment model MO2 may be larger than the weighting coefficient of the other items. By adopting such a configuration, it is possible to predict with higher accuracy whether or not a prescription of the triptan preparation TP is effective for a specific migraine patient.

[0065] In the information processing device 100 of this embodiment, the model acquisition unit 114 may input an image of a medical questionnaire IS including multiple medical questionnaire items, acquire a reading model MO1 which is a machine learning model that outputs answers by a patient to the multiple medical questionnaire items, and the target patient information acquisition unit 115 may acquire answers by the target patient to the multiple medical questionnaire items using the image of the medical questionnaire IS of the target patient and the reading model MO1. By adopting such a configuration, answers by the target patient to the multiple medical questionnaire items can be easily and accurately acquired.

[0066] B. Variations: The technology disclosed in this specification is not limited to the above-described embodiments, and can be modified in various forms without departing from the spirit of the invention. For example, the following modifications are also possible.

[0067] The configuration of the information processing device 100 in the above embodiment is merely an example and can be modified in various ways. Furthermore, the contents of the various processes in the above embodiment are merely an example and can be modified in various ways. For example, in the above embodiment, the information processing device 100 obtains the reading model MO1 and the efficacy judgment model MO2 by creating them, but the information processing device 100 may obtain the reading model MO1 and the efficacy judgment model MO2 generated by another device.

[0068] The interview items and machine learning algorithms used to create the headache diagnostic model MO in the above embodiment are merely examples and can be modified in various ways.

[0069] In the above-described embodiment, a part of the configuration realized by hardware may be replaced by software, and conversely, a part of the configuration realized by software may be replaced by hardware. [Explanation of symbols]

[0070] 100: Information processing device 110: Control unit 111: Prediction processing unit 112: Raw data acquisition unit 113: Learning data acquisition unit 114: Model acquisition unit 115: Target patient information acquisition unit 116: Prediction execution unit 120: Memory unit 130: Display unit 140: Operation input unit 150: Interface unit 190: Bus CP: Prediction processing program LD: Learning data MO1: Read model MO2: Drug efficacy judgment model MO: Headache diagnosis model RD: Prediction result data

Claims

1. An information processing device for predicting whether a headache patient is a specific migraine patient for whom a prescription of a triptan preparation is effective, a model acquisition unit that acquires a drug efficacy determination model, which is a machine learning model that receives answers to a plurality of interview items as input and outputs a determination result of whether or not a patient is the specific migraine patient; a target patient information acquisition unit for acquiring answers to the plurality of interview items for a target patient; a prediction execution unit that predicts whether the target patient is the specific migraine patient or not by using the answer of the target patient and the drug efficacy evaluation model, and outputs a prediction result; An information processing device comprising:

2. 2. The information processing device according to claim 1, An information processing device, wherein the plurality of interview items include items related to pain exacerbation / amelioration factors.

3. 3. The information processing device according to claim 2, The information processing device, wherein the plurality of medical interview items include at least one of an item regarding time of onset, an item regarding medical history and / or family history, and an item regarding pain itself.

4. 4. The information processing device according to claim 3, An information processing device, wherein in the drug efficacy assessment model, a weighting coefficient of an item relating to pain exacerbation / relief factors among the plurality of interview items is greater than weighting coefficients of other items.

5. The information processing device according to any one of claims 1 to 4, the model acquisition unit acquires a reading model, which is a machine learning model that receives an image of a medical questionnaire including the plurality of medical questionnaire items as an input and outputs answers to the plurality of medical questionnaire items; The target patient information acquisition unit is an information processing device that acquires answers to the plurality of medical questionnaire items targeted at the target patient using an image of the medical questionnaire of the target patient and the read model.

6. 1. An information processing method for predicting whether a headache patient is a specific migraine patient for whom prescription of a triptan preparation is effective, comprising: A step of acquiring a drug efficacy determination model, which is a machine learning model that receives answers to a plurality of interview items as input and outputs a determination result of whether or not a patient is the specific migraine patient; obtaining responses to the plurality of interview items from a target patient; predicting whether the subject patient is the specific migraine patient or not by using the answer of the subject patient and the drug efficacy evaluation model, and outputting the prediction result; An information processing method comprising:

7. A computer program for predicting whether a headache patient is a specific migraine patient for whom a prescription of a triptan preparation is effective, comprising: On the computer, A process of acquiring a drug efficacy determination model, which is a machine learning model that receives answers to a plurality of interview items as input and outputs a determination result of whether or not a patient is a specific migraine patient; A process of acquiring answers to the plurality of interview items for a target patient; A process of predicting whether the target patient is the specific migraine patient or not using the answer of the target patient and the drug efficacy evaluation model, and outputting the prediction result; A computer program that executes the following:

Images