Patient care plan determination system and patient care plan determination method

Abstract

A highly appropriate examination is recommended on the basis of the timing results of examination, diagnosis, treatment, and process observation in the overall workflow of patient management. This patient care plan determination system is provided with: a medical information server that holds at least attribute information and examination results of a plurality of patients; an information terminal that provides a patient management policy indicating the classification of the subject patient; and a computer that acquires predetermined information from the medical information server and the information terminal and determines a test item candidate relating to a disease of the subject patient. A computer executes: a process for estimating an examination value of a future examination item candidate for a subject patient by applying patient attribute information of the subject patient, an examination result of the subject patient, and a patient management policy of the subject patient to an examination plan learning completion model for calculating an examination value for each examination item; the examination value represents information which is corresponding to a patient management policy and can determine a disease; and a process for outputting the examination item candidates and the examination value of the subject patient (referring to Figure 6).

Description

Technical Field

[0001] The present disclosure relates to a patient care plan determination system and a patient care plan determination method. Background Art

[0002] The physician's role in ordering tests is to integrate and interpret a large amount of clinical information and select tests that are appropriate for each patient. Specifically, the physician (i) examines the patient and collects patient information. Based on the examination results, the physician lists suspected diseases and prioritizes the diseases. (ii) Considering the contribution to disease identification and the burden on the patient, the physician selects the required tests from all considered tests and performs them (issuing test orders). (iii) The physician interprets the test results and updates the list of suspected diseases. (iv) The physician repeats steps (ii) and (iii) until the disease is identified.

[0003] In order to accurately issue such examination instructions, doctors often need extensive experience. Therefore, for example, Patent Document 1 proposes a system that assists in the formulation of scheduled examinations and / or treatment plans for specific patients by displaying case information registered in a database on a screen.

[0004] Prior art literature

[0005] Patent Literature

[0006] Patent Document 1: Japanese Patent Application Laid-Open No. 2007-287027 Summary of the Invention

[0007] Problems to be solved by the invention

[0008] It is very difficult to implement appropriate examinations for patients based on the positioning and required results of examinations within the overall picture of patient management, including initial visits, follow-up visits, inpatient follow-up, and emergency treatment. Treatment and diagnosis of patients in hospitals are carried out according to management policies. Here, management policies refer to information that determines the examinations, diagnoses, and treatments to be implemented based on patient classification (patient positioning), such as initial visits, follow-up visits, inpatients, patients under ongoing observation, and emergency patients.

[0009] The system disclosed in Patent Document 1 can only reference past medical records registered in a database and cannot assist with examinations and treatments according to patient management guidelines. Specifically, even for the same symptoms, if the patient's location differs between initial and follow-up visits, the examination and treatment methods will differ. Therefore, appropriate examinations and treatments cannot be implemented based solely on past medical records. Therefore, if a system were available that considers the overall picture of patient management and recommends highly appropriate examination items at appropriate times, appropriate examinations could be implemented for various patients without relying on the physician's experience.

[0010] In view of such circumstances, the present disclosure proposes a technique for recommending highly appropriate examinations based on the results of each timing of examination, diagnosis, treatment, and process observation in the overall workflow of patient management.

[0011] Means for solving problems

[0012] In order to solve the above-mentioned problems, the present disclosure provides, as an example, a patient care plan determination system that determines at least candidate examination items related to a disease of a target patient.

[0013] The patient care plan decision system has:

[0014] a medical information server that stores at least attribute information and examination results of a plurality of patients;

[0015] an information terminal that provides a patient management policy indicating a classification of target patients; and

[0016] A computer that obtains predetermined information from a medical information server and an information terminal and determines candidate examination items related to a disease of a target patient.

[0017] Computer execution:

[0018] a process of acquiring attribute information of a target patient and a patient management policy of the target patient from a medical information server;

[0019] a process of estimating the examination value of a future candidate examination item for the subject patient by applying patient attribute information of the subject patient, the examination result of the subject patient, and the patient management policy of the subject patient to an examination plan learning completion model that calculates the examination value of each examination item, the examination value indicating information capable of identifying a disease according to the patient management policy; and

[0020] A process of outputting candidate examination items and examination values ​​for a target patient.

[0021] Further features related to the present disclosure will become apparent from the description of this specification and the accompanying drawings. In addition, the present disclosure is achieved and realized by the elements and combinations of various elements, the detailed description below, and the appended claims.

[0022] The descriptions in this specification are merely typical examples and do not limit the claims or application examples of the present disclosure in any sense.

[0023] Effects of the Invention

[0024] According to the technology disclosed herein, it is possible to recommend highly appropriate examinations based on the results of the respective timings of examinations, diagnoses, treatments, and process observations in the overall workflow of patient management. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 1 is a diagram showing a schematic configuration example of an information processing system (also referred to as a patient care plan determination system) 10 according to the present embodiment.

[0026] Figure 2 This is a diagram showing a functional configuration example (software configuration example) of the information processing system (patient care plan determination system) 10 according to the first embodiment.

[0027] Figure 3 This is a diagram for explaining the flow of data in the learning process of the inspection learning unit 300 of the inspection planning unit 1012 .

[0028] Figure 4 This is a flowchart for explaining the details of the learning parameter determination process according to the first embodiment.

[0029] Figure 5 This is a flowchart for explaining the details of the process of determining (learning) the threshold value of the inspection value when constructing the learned model.

[0030] Figure 6 This is a diagram for explaining the flow of data in the prediction process of the inspection prediction unit 600 of the inspection planning unit 1012 .

[0031] Figure 7 This is a flowchart for explaining details of the determination of examination item t+1 at time t+1 and the estimation of the probability of diagnosis of disease t+1 (diagnosis probability of disease t+1) in the first embodiment.

[0032] Figure 8 This is a diagram showing a configuration example of patient information 800 according to the present embodiment (common to the first and second embodiments).

[0033] Figure 9 This is a flowchart for explaining the inspection cost estimation process executed by the inspection planning unit 1012 .

[0034] Figure 10 This is a diagram showing a configuration example of a test instruction GUI 1000 displayed on the display screen of the information terminal 110 according to the first embodiment.

[0035] Figure 11 This is a diagram showing a functional configuration example (software configuration example) of an information processing system (patient care plan determination system) 10 according to the second embodiment.

[0036] Figure 12This is a diagram showing a detailed internal logical configuration example of the treatment planning unit 1100 and the examination planning unit 1012 in the patient care plan determination system 10 according to the second embodiment.

[0037] Figure 13 This is a diagram for explaining the flow of data in the learning process of the treatment learning unit t1301_1 to the treatment learning unit t+N1301_N of the treatment planning unit 1100.

[0038] Figure 14 This is a flowchart for explaining the learning parameter determination process (learning completion model generation process) of the medical treatment learning unit t1301_1 to the medical treatment learning unit t+N_1301_N in the second embodiment.

[0039] Figure 15 This is a diagram for explaining the flow of data in the treatment prediction process of the treatment prediction unit 1500 of the treatment planning unit 1100 .

[0040] Figure 16 This is a flowchart for explaining the processing of multiple patterns of patient management policies, diagnosis and treatment methods, and patient disease information from the estimated time point t+1 (time step t+1) to the time point t+N (time step t+N) in the second embodiment.

[0041] Figure 17 This is a diagram for explaining the flow of data in the learning process of the inspection learning unit 1700 of the inspection planning unit 1012 according to the second embodiment.

[0042] Figure 18 This is a flowchart for explaining the details of the learning parameter determination process of the second embodiment.

[0043] Figure 19 This is a diagram showing a configuration example of an output screen (UI: User Interface) 1900 for a treatment plan and an examination plan according to the second embodiment.

[0044] Figure 20 This is a diagram showing a configuration example of an inspection instruction GUI 2000 displayed on the display screen of the information terminal 110 according to the second embodiment. DETAILED DESCRIPTION

[0045] This embodiment discloses an information processing system (patient care plan determination system) that predicts a treatment plan including various processes and results of patient management stages, examinations, diagnoses, treatments, and process observations based on patient information, and infers the examination items required to execute the treatment plan.

[0046] The following describes embodiments of the present disclosure with reference to the accompanying drawings. Functionally identical elements are sometimes shown with the same reference numerals in the accompanying drawings. Furthermore, the accompanying drawings illustrate specific embodiments and installation examples based on the principles of the present disclosure. However, these are intended for understanding the present disclosure and are not intended to limit the present disclosure in any way.

[0047] In the present embodiment, those skilled in the art have described it in sufficient detail for the purpose of implementing the present disclosure. However, other installations and methods are also possible, and it should be understood that changes in structure and construction and replacement of various elements can be made without departing from the scope and spirit of the technical concept of the present disclosure. Therefore, the following description is not to be interpreted as being limited to this.

[0048] (1) First embodiment

[0049] <Hardware Configuration Example of Information Processing System>

[0050] Figure 1 This figure shows a schematic configuration example of an information processing system (also referred to as a patient care plan determination system) 10 according to this embodiment. The hardware configuration example of the information processing system 10 is common to the first embodiment and the second embodiment described later.

[0051] The information processing system (patient care plan determination system) 10 includes, for example, a computer 100 installed on the cloud, an information terminal 110 installed in a doctor's examination room, and an external recording device (medical information server) 111 , which are connected via a network (intra-hospital network) 109 .

[0052] Computer 100 includes a processor 101, such as a CPU, a main storage device 102, a secondary storage device 103, a network adapter 104, an input device 105, and an output device 106. The main storage device 102 stores, for example, a program for determining examination instructions (described later). The secondary storage device 103 stores, for example, information obtained from an external recording device 111 (electronic medical records, receipt information, guidelines, etc.). The processor 101 reads the program from the main storage device 102, expands it into internal memory (not shown), and implements the various processing units (various functions for determining examination instructions) described later.

[0053] The input device 105 is comprised of, for example, a keyboard, mouse, or touch panel, for inputting information into the computer 100. The output device 106 is comprised of, for example, a display, printer, or other device for outputting information. The network adapter 104 receives information from the information terminal 110 via the network 109 and transmits it to the processor 101 or stores it in the secondary storage device 103. Furthermore, the network adapter 104 transmits, for example, determined inspection instruction information to the information terminal 110 via the network 109 in response to instructions from the processor 101.

[0054] <Functional Configuration Example of Information Processing System>

[0055] Figure 2 This is a diagram showing a functional configuration example (software configuration example) of the information processing system (patient care plan determination system) 10 according to the first embodiment.

[0056] The information processing system 10 functionally includes an examination sorting system 1110 provided in an external recording device (medical information server) 111 , a patient information acquisition unit 1011 and an examination planning unit 1012 provided in a processor 101 in a computer 100 , and an examination instruction output unit 1101 provided in an information terminal 110 .

[0057] The information processing system 10 also processes electronic medical record information 1111, receipt information 1112, and guideline information 1113 stored in an external recording device (medical information server) 111, as well as patient information 201, patient management policy 202, and future examination item candidates 203 and examination value 204. Receipt information 1112 also includes medical treatment fee (points) information. Guideline information 1113 stores information on medical examination items determined based on symptoms (e.g., performing examination X for symptoms A and B). Furthermore, patient management policy 202 contains information indicating patient classifications, such as initial visit, follow-up visit, hospitalization, ongoing observation, and emergency treatment.

[0058] The patient information acquisition unit 1011 accesses the electronic medical record information 1111 from the external recording device (medical information server) 111, obtains information on past examinations and diagnoses corresponding to the target patient, and obtains this information as patient information 201. Furthermore, the patient information acquisition unit 1011 obtains (receives) and outputs the patient management policy 202 entered (selected) by the physician from the information terminal 110. Patient information 201 includes, for example, patient attribute information (gender, age, etc.), previously performed examinations and examination result information, and patient disease information identified in previous examinations. In this specification, examination result information at time point t (time step t) is referred to as examination result information t, and examination result information at time point t+1 (time step t+1) is referred to as examination result information t+1. Furthermore, disease information identified by examination t+1 performed at time point t+1 (time step t+1) is referred to as patient disease information t+1. If a disease is not identified by examination t at time point t, the disease is identified by examination t+1.

[0059] The examination planning unit 1012 uses the patient information 201 and the patient management policy to calculate an examination value 204 representing the amount of information corresponding to the disease category taking the patient management policy into consideration (the probability of the disease being identified based on each examination item), and identifies examination items with a large amount of information (high probability values) as future examination item candidates 203. The examination planning unit 1012 transmits the identified future examination item candidates 203 to the information terminal 110 as recommended examination information.

[0060] The examination order output unit 1101 receives the future examination item candidates 203 and displays them on a display screen (not shown) of the information terminal 110. When the doctor confirms the future examination item candidates 203 displayed on the display screen and selects a desired examination item from the candidates, the examination order output unit 1013 transmits the examination order to the examination ranking system 1110.

[0061] The inspection sequencing system 1110 executes the inspection instructions received from the information terminal 110 on the inspection room.

[0062] <Data Flow of Learning Processing by the Inspection Learning Unit 300 of the Inspection Planning Unit 1012>

[0063] Figure 3 This is a diagram for explaining the data flow in the learning process of the inspection learning unit 300 of the inspection planning unit 1012. The inspection planning unit 1012 includes: the inspection learning unit 300, which generates a learning completion model; and the inspection prediction unit 400, which uses the learning completion model to infer inspection items, etc. Figure 3 In the example, the inspection learning unit 300 generates various data and outputs a learning-completed model 310 generated based on the data.

[0064] When generating a learned model, the inspection and learning unit 300 obtains as input data the following: criterion information, patient management policy t at time point t (time step t), patient attribute information, examination items, examination result information t at time point t, examination result information t+1 at time point t+1 (time step t+1), and patient disease information t+1 at time point t+1. Criteria information and patient attribute information are fixed when generating the learned model, but other information changes at each time point. Specifically, the patient management policy t at a certain time point t (for example, if the patient had their first visit last time, this time is their second visit, so the management policy changes), the examination result at that time point t, the examination result information t+1 at the next time point t+1, and the patient disease information t+1 determined based on the examination result information t+1, all change depending on the time. The inspection and learning unit 300 performs learning using information from multiple time points in this manner to construct the learned model 310.

[0065] The inspection learning unit 300 uses a probability distribution model to collect inspection result data for each inspection item and estimate a probability distribution 301 (estimation of the probability distribution 301 in the inspection result). As the number of data increases, the probability distribution 301 approaches a normal distribution.

[0066] Furthermore, the inspection learning unit 300 generates a probability distribution function through learning based on the estimated probability distribution, and generates a restored inspection result 302 using the probability distribution function.

[0067] Next, the inspection learning unit 300 compares the restored inspection result 302 with the input inspection result to estimate the inspection value 303 of each inspection item.

[0068] Furthermore, the inspection learning unit 300 estimates a learning parameter 304 for estimating a threshold value of inspection value, that is, a threshold value for determining the inspection value of an inspection item, and stores the learning parameter threshold value 304 in the primary storage device 102 or the secondary storage device 103 according to the inspection item.

[0069] Then, the examination learning unit 300 estimates the diagnosis probability for the disease information t+1 and the risk of the disease at time t+1 (disease diagnosis probability and risk 305 ).

[0070] The examination learning unit 300 determines learning parameters and outputs them as a learned model 310. The learning parameters include, for example, parameters of a neural network for extracting features from patient information, parameters for estimating probability distribution, parameters for restoring examination results, and parameters for estimating examination value.

[0071] <Details of Learning Parameter Determination Process>

[0072] Figure 4 This is a flowchart for describing the details of the learning parameter determination process of the first embodiment. In the following description, the inspection learning unit 300 is the main body of each step. However, since the inspection planning unit 1012, which includes the inspection learning unit 300, is implemented by developing a program in the processor 101, the processor 101 may also be the main body of the operation.

[0073] (i)S401

[0074] The examination and learning unit 300 receives the patient information required to generate the learned model 310. To generate a more accurate learned model 310, it is necessary to obtain as much patient information as possible. The patient information comprises various pieces of information about various patients, including the aforementioned guideline information, patient management policy t at time point t (time step t), patient information, examination items, examination result information t at time point t, examination result information t+1 at time point t+1 (time step t+1), and patient disease information t+1 at time point t+1.

[0075] (ii)S402

[0076] The examination and learning unit 300 integrates the patient information obtained in S401 and extracts the characteristic values ​​of the patient information. Specifically, the examination and learning unit 300 converts the received patient information into text and digitizes it in a manner that can be processed by a neural network (integration of the patient information), applies the integrated patient information to the neural network (e.g., deep learning), and thereby calculates the characteristic values ​​of the numerical information.

[0077] (iii)S403

[0078] The inspection learning unit 300 collects a plurality of inspection results for each inspection item and estimates a probability distribution (function) for each inspection item using a probability distribution model.

[0079] (iv)S404

[0080] The inspection learning unit 300 restores the inspection result based on the probability distribution estimated in S403. This is because the inspection result as the correct value is already known, and therefore it is a process to confirm whether the correct value is obtained based on the estimated probability distribution.

[0081] (v)S405

[0082] The examination learning unit 300 estimates the examination value based on the examination results restored in S404. If the original examination results can be restored based on the estimated probability distribution, the estimated probability distribution is correctly estimated. In S405, the examination value is estimated to what extent if new patient information is input.

[0083] (vi)S406

[0084] The inspection learning unit 300 estimates a learning error in the inspection value estimated in S405. Here, the learning error refers to the error between the original inspection result and the restored inspection result.

[0085] (vii)S407

[0086] The learning inspection unit 300 determines whether the learning error value estimated in S406 has converged. Convergence is determined by whether it is less than a preset threshold. If the learning error value has converged (if yes in S407), the process moves to S408. If the learning error value has not converged (if no in S407), the process moves to S409.

[0087] When the learning error is small, a learned model can be constructed that can predict the patient's examination value when new patient information is input.

[0088] (viii)S408

[0089] The inspection learning unit 300 sets learning parameters for convergence of learning errors in the learned model. The learning parameters include the parameters of the neural network in S402, parameters used for probability distribution estimation, parameters used for inspection result restoration, and parameters used for inspection value estimation.

[0090] (ix)S409

[0091] The inspection learning section 300 updates learning parameters (for example, parameters of a neural network) so that the learning error decreases according to the optimization function used for updating.

[0092] <Details of Threshold Determination Process>

[0093] Figure 5 This is a flowchart for explaining the details of the process of determining (learning) the threshold value of the inspection value when constructing the learned model.

[0094] (i)S501

[0095] The inspection learning unit 300 obtains the inspection value and criterion information 1113 for each inspection item.

[0096] (ii)S502

[0097] The inspection learning unit 300 receives inspection item t+1 at time t+1 (time step t+1).

[0098] (iii)S503

[0099] The inspection learning unit 300 estimates a threshold value for determining the inspection value of an inspection item. The initial value (fixed value) of the threshold value can be predetermined, and learning parameters for estimating the threshold value of the inspection value can be determined while changing the parameters to reduce the learning error.

[0100] This threshold is assumed to vary depending on the management policy (patient classification: initial visit, follow-up visit, hospitalized, under observation, emergency, etc.), so the threshold for the examination value is estimated in S503. The threshold is estimated according to the examination item, and in subsequent processing, learning is performed in a manner consistent with the management policy, thereby determining (estimating) the final threshold for the examination value.

[0101] (iv)S504

[0102] The inspection learning unit 300 estimates the diagnosis probability and risk (the level of risk of the diagnosed disease itself) of the disease information t+1 obtained from the inspection result t+1 corresponding to the inspection item t+1. The threshold estimated in S503 may not be accurate, but the threshold is used based on the Figure 4 The diagnosis probability and risk of a specific disease are estimated based on the test results restored from the estimated probability distribution and the estimated threshold value during the processing.

[0103] (v)S505

[0104] The examination learning unit 300 estimates the management policy t at time point t (time step t) based on the diagnostic probability and risk information estimated in S504. That is, in S505, it estimates which category the patient belongs to (initial visit, return visit, hospitalized, under observation, emergency, etc.).

[0105] (vi)S506

[0106] The examination learning unit 300 acquires the actual patient management policy t.

[0107] (vii)S507

[0108] The examination learning unit 300 compares the patient management policy t estimated in S505 with the actual patient management policy t acquired in S506 to estimate a learning error.

[0109] (viii)S508

[0110] The learning inspection unit 300 determines whether the learning error value estimated in S507 has converged. Convergence is determined by whether it is less than a preset threshold. If the learning error value has converged (if yes in S508), the process proceeds to S509. If the learning error value has not converged (if no in S508), the process proceeds to S510.

[0111] (ix)S509

[0112] The inspection learning unit 300 determines the parameter of the inspection value threshold value estimated in S503 as the learning parameter of the estimated inspection value threshold value.

[0113] (x)S510

[0114] The inspection learning unit 300 updates the learning parameters (neural network parameters) for estimating the threshold value of the inspection value according to the optimization function. Then, using the updated learning parameters, the process of estimating the threshold value of the inspection value of the determined inspection item is executed again in S503.

[0115] <Data Flow of Prediction Processing by the Inspection Prediction Unit 600 of the Inspection Planning Unit 1012>

[0116] Figure 6 This is a diagram for explaining the data flow in the prediction process of the inspection prediction unit 600 of the inspection planning unit 1012. Figure 6 In the process, the inspection prediction unit 600 applies the data at time point t (time step t) to the learning completion model, and outputs the inspection item t+1 at time point t+1 (time step t+1) and the diagnosis probability and risk of disease t+1 at time point t+1.

[0117] When predicting examination item t+1, for example, the examination prediction unit 600 receives as input the patient management policy t, patient attribute information, examination items, and examination result t at time t. For example, based on the most recent patient management policy and examination results, the unit predicts the examination items to be performed at the next time point, the probability of being diagnosed with a specific disease, and the risk level of the disease.

[0118] The examination prediction unit 600 applies the patient management policy t, patient attribute information, and examination item and examination result t at time t to the learned model generated by the examination learning unit 300. This generates a candidate for examination item t+1 at time t+1, its examination result 601, an examination value 602 for each candidate for examination item t+1, and a threshold 603 for the examination value. As described above, for example, examination item t can be set as the most recent examination item, examination result t can be set as the most recent examination result corresponding to the examination item, and the candidate for examination item t+1 can be set as the candidate examination item to be accepted at the next time.

[0119] The inspection prediction unit 600 also determines whether the inspection value of the inspection item t+1 candidate (the inspection value of each inspection item candidate) 602 is greater than the threshold 603 in the inspection value, determines the inspection item t+1 that is greater than the threshold, and infers the probability (diagnosis probability) of being diagnosed as the disease at the t+1 time point (the disease at the next time point).

[0120] <Details of the decision (forecast) on inspection items, etc.>

[0121] Figure 7This flowchart details the determination of examination item t+1 at time point t+1 (time step t+1) and the estimation of the probability of being diagnosed with disease t+1 (the diagnosis probability of disease t+1) in the first embodiment. In the following description, the examination prediction unit 600 is the primary operator of each step. However, since the examination planning unit 1012, which includes the examination prediction unit 600, is implemented by developing a program in the processor 101, the processor 101 may also be the primary operator.

[0122] (i)S701

[0123] The examination prediction unit 600 receives (acquires) information on a patient to be processed, the learned model 310 , and the patient management policy t, input by an operator, for example.

[0124] (ii)S702

[0125] The examination prediction unit 600 integrates the patient information obtained in S701 and extracts the characteristic values ​​of the patient information. Specifically, the examination prediction unit 600 converts the received patient information into text and digitizes it in a manner that can be processed by a neural network (integration of the patient information), applies the integrated patient information to the neural network (e.g., deep learning), and thereby calculates the characteristic values ​​of the numerical information.

[0126] (iii)S703

[0127] The inspection prediction unit 600 applies the feature amount extracted in S702 for each candidate of inspection item t+1 to the learned model 310 for a plurality of inspection items, thereby estimating the inspection result.

[0128] (iv)S704

[0129] The examination prediction unit 600 quantifies the examination results estimated in S703 to estimate the examination value of each candidate for examination item t+1. The examination value is a probability value indicating how much the diagnosis probability of a disease would increase if the corresponding examination item candidate were selected. The examination value is expressed as information content and is obtained through calculations based on the learning completion model 310.

[0130] (v)S705

[0131] The inspection prediction unit 600 specifies the inspection item having the highest value among the inspection values ​​estimated in S704 .

[0132] (vi)S706

[0133] The inspection prediction unit 600 estimates a threshold value for determining the inspection value based on the management policy t and the amount of information on the inspection value.

[0134] (vii)S707

[0135] The inspection prediction unit 600 determines whether the inspection value determined in S705 is greater than the threshold value estimated in S706. If the inspection value is greater than the threshold value (if yes in S707), the process proceeds to S708. If the inspection value is less than the threshold value (if no in S707), the process proceeds to S710.

[0136] (viii)S708

[0137] The inspection prediction unit 600 determines the inspection item with the highest inspection value as inspection item t+1_610.

[0138] (ix)S709

[0139] The examination prediction unit 600 estimates the probability and risk 620 of being diagnosed with disease t+1 when examination item t+1 is performed.

[0140] Specifically, the inspection prediction unit 600 restores the inspection value and estimates the disease diagnosis probability based on the inspection value probability distribution information of the learning completion model 310. In addition, the inspection prediction unit 600 estimates the inspection value based on the estimated disease diagnosis probability. Then, the inspection prediction unit 600 determines the inspection item with the greatest inspection value (disease diagnosis probability). The disease diagnosis probability corresponding to the determined inspection item becomes the probability of being diagnosed with the above-mentioned disease t+1. In this way, the estimation of the inspection value is carried out based on the disease probability estimation, so the disease probability will also be estimated during the inspection value estimation process. In addition, the inspection prediction unit 600 refers to the criterion information and determines the risk (disease risk) 620 according to the rules.

[0141] (x)S710

[0142] The inspection prediction unit 600 adds the estimated inspection results of the inspection items specified in S705 to the inspection result t.

[0143] (xi)S711

[0144] The inspection prediction unit 600 updates the candidates for inspection item t+1. For example, if the number of inspection item candidates is set to every k (m is a positive integer), the next group of m inspection item candidates is set as the new inspection item to be processed. Then, the processing from S703 onwards is performed on the new inspection item candidates.

[0145] <Example of Patient Information>

[0146] Figure 8 This is a diagram showing a configuration example of patient information 800 according to the present embodiment (common to the first and second embodiments).

[0147] The patient information includes identification information (ID) 801 , patient attribute information 802 , date and time 803 , patient management policy 804 , a result group 805 of a plurality of examinations (examination items A to Z), and disease information 806 as structural items.

[0148] Identification information (ID) 801 is information for uniquely identifying and recognizing a patient. Patient attribute information 802 is information indicating the patient's gender and age. Date and time 803 is information indicating the date and time of the examination. As described above, the patient management policy 804 is information indicating the classification of patients such as initial visit, follow-up visit, hospitalization, ongoing observation, and emergency treatment. The result group 805 of multiple examinations (examination items A to Z) is information indicating the examination values ​​and judgment results (abnormal (the value is too high, so it is abnormal (H), or the value is too low, so it is abnormal (L)), normal) corresponding to each examination item. Disease information 806 is information indicating the type of disease diagnosed by the doctor based on the examination results.

[0149] <Inspection Cost Estimation Processing>

[0150] Figure 9 This is a flowchart for explaining the inspection cost estimation process executed by the inspection planning unit 1012. The inspection cost estimation process is a process based on rules and does not involve learning.

[0151] (i)S901

[0152] The examination planning unit 1012 receives receipt information 1112 including information on medical treatment reimbursement calculation conditions.

[0153] (ii)S902

[0154] The inspection planning unit 1012 receives information on the inspection item t+1_610 predicted by the inspection prediction unit 600 .

[0155] (iii)S903

[0156] The examination planning unit 1012 receives the patient management policy t of the target patient.

[0157] (iv)S904

[0158] The examination planning unit 1012 estimates (calculates) the medical treatment remuneration score for the examination item t+1_610 based on the medical treatment remuneration calculation conditions acquired in S901.

[0159] (v)S905

[0160] The examination planning unit 1012 estimates (calculates) the hospital cost burden based on the medical treatment reimbursement points estimated (calculated) in S904. The information on the estimated hospital cost burden is sent to the information terminal 110 and can be displayed on the display screen of the information terminal 110.

[0161] <Configuration Example of GUI Displayed on the Display Screen of Information Terminal 110>

[0162] Figure 10 This figure shows a configuration example of a test order GUI 1000 displayed on the display screen of the information terminal 110 according to the first embodiment. Through the test order GUI 1000, patient management policies are selected and displayed (doctor selection), and disease determination results and test items are output.

[0163] The examination instruction GUI 1000 includes, for example, a management policy selection / display unit 1001 , a disease probability display unit 1002 , and a disease examination item display unit 1003 as components.

[0164] When receiving recommended examination information from the computer 100 of the patient care plan determination system 10 , the management policy selection / display unit 1001 displays the patient management policy of the target patient that can be selected by the doctor and the currently selected patient management policy.

[0165] The disease probability display unit 1002 provides information on the disease name, disease probability, and disease risk (the degree of risk of the disease itself) obtained through examinations and disease prediction processing (prediction simulation). For example, if the prediction results indicate that disease A is present with a 70% probability, the risk of the disease is moderate.

[0166] The disease examination item display unit 1003 displays, for example, examination items for diseases with a disease probability higher than a predetermined value or examination items for high-risk diseases, patient costs indicating the fees paid by the patient for the examination, and hospital burden indicating the cost burden on the hospital side. Figure 10 In the example, the examination items for disease A with a probability of 70% and disease C with a high risk are displayed.

[0167] (2) Second embodiment

[0168] The second embodiment further proposes a patient care plan determination system 10 that provides a diagnosis and treatment plan for a target patient based on the examination plan of the first embodiment. The hardware structure of the patient care plan determination system 10 of the second embodiment is as follows: Figure 1 As shown, therefore, its description is omitted.

[0169] <Functional Configuration Example of Information Processing System>

[0170] Figure 11 This is a diagram showing a functional configuration example (software configuration example) of an information processing system (patient care plan determination system) 10 according to the second embodiment.

[0171] The information processing system 10 functionally includes an examination sorting system 1110 provided in an external recording device (medical information server) 111, a patient information acquisition unit 1011 provided in a processor 101 in a computer 100, a diagnosis and treatment planning unit 1100 and an examination planning unit 1012, and an information terminal 110 and an examination instruction output unit 1101.

[0172] The information processing system 10 also processes electronic medical record information 1111, receipt information 1112, and guideline information 1113 stored in an external recording device (medical information server) 111, as well as patient information 201, patient management policy 202, and examination plan i_10121 (including, for example, the aforementioned candidate future examination items 203 and examination value 204). Receipt information 1112 also includes medical treatment remuneration (points) information. Guideline information 1113 stores information on medical examination items determined based on symptoms (e.g., performing examination X for symptoms A and B).

[0173] The patient information acquisition unit 1011, similar to the first embodiment, accesses the electronic medical record information 1111 of the external recording device (medical information server) 111, acquires the information of the past examinations and past diagnoses corresponding to the target patient, and acquires them as the patient information 201. In addition, the patient information acquisition unit 1011 acquires (receives) the patient management policy 202 input (selected) by the doctor from the information terminal 110 and outputs it. The patient information 201 is Figure 8 The information shown is composed of

[0174] The treatment planning unit 1100 uses the patient information 201 and the patient management policy 202 to predict (estimate) future management policies, treatment methods, and disease information, and outputs the predicted information as a patient treatment plan 11001. For example, if the patient is an inpatient, the patient treatment plan 11001 includes information such as whether the current treatment will continue in the future and, if a treatment method is changed, which method will be used.

[0175] The examination plan unit 1012 determines (predicts) an examination plan 10121 using the patient information 201 and the patient management policy 202, as in the first embodiment. As in the first embodiment, the examination plan 10121 includes information on future examination items, examination results, and a prediction of the examination value. The examination plan 10121 includes future examination item candidates 203 and examination value 204 (see Figure 2The examination planning unit 1012 calculates the examination value 204, which represents the amount of information corresponding to the disease category (the probability of the disease being determined based on each examination item), taking into account the patient management policy 202, and identifies examination items with a large amount of information (high probability value) as future examination item candidates 203. The examination planning unit 1012 transmits the identified future examination item candidates 203 as recommended examination information to the information terminal 110.

[0176] The examination order output unit 1013 receives the future examination item candidates 203 and displays them on a display screen (not shown). When the doctor confirms the future examination item candidates 203 displayed on the display screen and selects a desired examination item from the candidates, the examination order output unit 1013 sends the examination order to the examination ranking system 1110.

[0177] The inspection sequencing system 1110 executes the inspection instructions received from the information terminal 110 on the inspection room.

[0178] <Detailed Internal Logical Structure Example of the Treatment Planning Unit 1100 and the Examination Planning Unit 1012>

[0179] Figure 12 This is a diagram showing a detailed internal logical configuration example of the treatment planning unit 1100 and the examination planning unit 1012 in the patient care plan determination system 10 according to the second embodiment.

[0180] (i) The treatment planning unit 1100 has N+1 treatment prediction units, namely, the treatment prediction unit t_1100_1 at time point t (time step t), the treatment prediction unit t+1_1100_2 at time point t+1 (time step t+1), the treatment prediction unit t+2_1100_3 at time point t+2 (time step t+2), and the treatment prediction unit t+1_1100_N+1 at time point t+N (time step t+N).

[0181] 1 diagnosis and treatment prediction unit t+k_1100_k+1 (k=0, 1, ..., N) obtains (accepts) the patient management policy and patient information from time point tk (time step tk) to time point t (time step t) (refer to Figure 8 )(obtain information before diagnosis and treatment prediction), use this information to generate diagnosis and treatment plan prediction information consisting of patient management policy t+k, diagnosis and treatment method t+k and patient disease information t+k at time point t+k (time step t+k), and output it to the inspection planning unit 1012.

[0182] (ii) The inspection planning unit 1012 has N+1 inspection prediction units, namely, the inspection prediction unit t_1012_1 at time point t (time step t), the inspection prediction unit t+1_1012_2 at time point t+1 (time step t+1), the inspection prediction unit t+2_1012_3 at time point t+2 (time step t+2), and the inspection prediction unit t+1_1012_N+1 at time point t+N (time step t+N).

[0183] The examination prediction unit t_1012_1 obtains the patient information from the medical information server 111 ( Figure 8 ), examination items and examination result information t, and similarly to the first embodiment, estimate the probability distribution in the examination results by examination item → restore the examination results at the next time step t+1 → estimate the examination value, disease diagnosis probability and its risk. The examination prediction unit t+k_1012_k+1 (k=0, 1, ..., N) obtains patient information, examination items and examination result information at time point t+k (time step t+k) (the examination prediction result estimated by the examination prediction unit t+k-1) t+k, estimates the probability distribution in the examination results by examination item → restore the examination results at the next time step t+k+1 → estimate the examination value, disease diagnosis probability and its risk, and outputs this information.

[0184] (iii) In addition, when the examination is actually carried out, the examination results 1200_1 of the patient at time point t (time step t), the examination results 1200_2 of the patient at time point t+1 (time step t+1), the examination results 1200_3 of the patient at time point t+2 (time step t+2), ..., the examination results 1200_N+1 of the patient at time point t+N (time step t+N) are output and collected and stored in the medical information server 111 as examination results t, ..., t+N_1201.

[0185] <Data Flow of Learning Processing by the Treatment Learning Unit 1301 of the Treatment Planning Unit 1100>

[0186] Figure 13 This is a diagram for explaining the data flow in the learning process of the treatment learning unit t1301_1 to the treatment learning unit t+N1301_N of the treatment planning unit 1100. The treatment planning unit 1100 includes the treatment learning units t1301_1 to the treatment learning units t+N1301_N that generate the learning completion model, and the treatment prediction unit 1500 (see Figure 15 ).exist Figure 13 In the process, the diagnosis and treatment learning unit t1301_1 to the diagnosis and treatment learning unit t+N1301_N generate various data and output the learning completed model 1303 generated based on these data.

[0187] The diagnosis and treatment learning unit t1301_1 obtains the patient management policy at time point t (time step t), the diagnosis and treatment method information t at time point t (time step t), the patient attribute information, the examination items, the examination result information t at time point t (time step t), and the patient disease information t at time point t (time step t) as input data. Patient attribute information and examination items are fixed information when generating the learning completion model, but other information changes at each time point. That is, the patient management policy t at a certain time point t (time step t) (for example, if the patient was diagnosed for the first time last time, this is the second time for diagnosis, so the management policy changes), the examination result at that time point t (time step t), the examination result information t+1 at the next time point t+1 (time step t+1), and the patient disease information t+1 determined based on the examination result information t+1 become information that changes according to the time. The diagnosis and treatment learning units 1301_1 to 1301_N use information from multiple time points in this way to perform learning in sequence and construct the learning completion model 1303.

[0188] The diagnosis and treatment learning unit t1301 at time point t (time step t) and the diagnosis and treatment learning unit t+k1301_k at time point t+k (k=1,…,N) (time step t+k) use patient attribute information, patient management policies t and t+k at time point t and time point t+k (time step t and time step t+k), and diagnosis and treatment method information t+k to infer the patient management policy t+k+1, diagnosis and treatment method t+k+1, and patient disease information t+k+1 at time point t+k+1 (time step t+k+1). The diagnosis and treatment learning unit t+k1301_k then compares the actual patient management policy t+k+1, the implemented diagnosis and treatment method t+k+1, and the actual patient disease information t+k+1 with the predicted (estimated) patient management policy t+k+1, the predicted diagnosis and treatment method t+k+1, and the predicted patient disease information t+k+1, and estimates the error t+k1302_k of the learned model. Furthermore, the diagnosis and treatment learning unit t+k1301_k repeatedly performs the learning process to converge the error t+k of the learned model and determines the learning parameters of the learned model 1303.

[0189] <Learning Completion Model Generation Processing of the Medical Learning Department>

[0190] Figure 14This is a flowchart for explaining the learning parameter determination process (learning completion model generation process) of the diagnosis and treatment learning unit t1301_1 to the diagnosis and treatment learning unit t+N_1301_N in the second embodiment. In the following description, the action subject of each step is set to the diagnosis and treatment learning unit, but this is a general term for the diagnosis and treatment learning unit t1301_1 or the diagnosis and treatment learning unit t+k1301_k (k is an integer from 1 to N). In addition, the diagnosis and treatment planning unit 1100 including the diagnosis and treatment learning unit 1301 is implemented by expanding the program in the processor 101, so the processor 101 can also be used as the action subject.

[0191] (i)S1401

[0192] The diagnosis and treatment learning unit 1301 obtains information of multiple patients from the medical information server 111, i.e., patient information and patient management policies of the multiple patients from time point tk to time point t (time step tk to time step t). Here, the patient information from time step tk to t includes patient attribute information, examination items and examination result information from time point tk to time point t (time step tk to t), patient disease information, and diagnosis and treatment method information (for the structural items of patient information, refer to Figure 8 ).

[0193] (ii)S1402

[0194] The diagnosis and treatment learning unit 1301 integrates the patient information and patient management policies obtained in S1401 and extracts the features of this information. Specifically, the diagnosis and treatment learning unit 1301 converts the received patient information and patient management information into text, converts them into digits in a manner that can be processed by a neural network (information integration), and applies the integrated information to the neural network (e.g., deep learning) to extract the features of the digitized information.

[0195] (iii)S1403

[0196] The diagnosis and treatment learning unit 1301 uses the feature quantity extracted in S1402 to infer the patient management policy, diagnosis and treatment method, and patient disease information from time point t+1 to time point t+N (time step t+1 to time step t+N). For example, it can be inferred by machine learning based on neural networks such as deep learning. For example, by using RNN (Recurrent Neural Network) and the like to perform time series learning on feature quantities such as patient information and management policies at various time points in the past, it is possible to predict patient management policies, diagnosis and treatment methods, and patient disease information at various time points in the future. More specifically, the machine learning model used in the prediction operation uses patient information and patient management policies at various time points in the past as time series observation data, and predicts changes in patient disease information as the therapeutic effect of a specific diagnosis and treatment method.

[0197] (iv)S1404

[0198] The diagnosis and treatment learning unit 1301 compares the patient management policy, diagnosis and treatment method and patient disease information (diagnosis and treatment plan) from time step t+1 to time step t+N obtained in S1403 with the actual patient management policy, diagnosis and treatment method and patient disease information from time step t+1 to time step t+N to estimate the learning error.

[0199] Here, the learning error represents the error of the estimated result associated with the treatment plan mode 1 to mode M (M is an integer greater than or equal to 2). The learning error can be a weighted average of the actual error and the virtual error. The actual error, for example, represents the difference from the patient information obtained by the treatment plan actually performed on patient A (for example, treatment plan 1). On the other hand, the virtual error represents the difference from the virtual patient information obtained by the treatment plan that was not actually performed (for example, treatment plan 2). The virtual patient information is obtained from other patients (for example, patient B) who actually performed treatment plan 2. In addition, patient B is selected from a group of patients whose feature values ​​of patient information are similar to those of patient A.

[0200] (v)S1405

[0201] The diagnosis and treatment learning unit 1301 compares the learning error estimated in S1404 with a preset threshold to determine whether the learning error has converged. If the learning error has converged (if yes in S1405), the process moves to S1406. On the other hand, if the learning error has not converged (if no in S1405), the process moves to S1407.

[0202] (vi)S1406

[0203] The medical treatment learning unit 1301 determines the parameters of the learned model used in the feature value extraction in S1402 and the estimation of the patient management policy, medical treatment method, and patient disease information in S1403 as the parameters of the learned model.

[0204] (vii)S1407

[0205] The medical treatment learning unit 1301 updates the learning parameters according to the optimization function so as to reduce the learning error. The updated learning parameters are used when re-estimating information such as the patient management policy in the iterative process.

[0206] <Data Flow of Prediction Processing by the Treatment Prediction Unit 1500 of the Treatment Planning Unit 1100>

[0207] Figure 15 This is a diagram for explaining the data flow in the treatment prediction process of the treatment prediction unit 1500 of the treatment planning unit 1100. The treatment prediction unit 1500 applies input data to the learned model 1303 to predict tuberculosis.

[0208] The diagnosis and treatment prediction unit 1500 obtains as input data the patient management policy tk at time point tk (time step tk) to the patient management policy t at time point tk (time step t), diagnosis and treatment method information tk at time point tk (time step tk) to the diagnosis and treatment method information t at time point tk (time step tk), patient attribute information, examination items and examination result information tk at time point tk (time step tk) to the examination result information t at time point tk (time step t), and patient disease information tk at time point tk (time step tk) to the patient disease information t at time point tk (time step tk). In the second embodiment, the information of the patient from time point tk (time step tk) to time point t (time step t) of the treatment target is used to predict the diagnosis and treatment plan t+1 at time point t+1 (time step t+1) to the diagnosis and treatment plan t+N at time point t+N (time step t+N).

[0209] The diagnosis and treatment prediction unit 1500 applies the patient management policy tk at time point tk (time step tk) to the patient management policy t at time point t (time step t), the diagnosis and treatment method information tk at time point tk (time step tk) to the diagnosis and treatment method information t at time point tk (time step tk), the patient attribute information, the examination items and the examination result information tk at time point tk (time step tk) to the examination result information t at time point tk (time step t), the patient disease information tk at time point tk (time step tk) to the patient disease information t at time point tk (time step t) to the learning completion model 1303, and calculates the predicted patient management policy t+1, the predicted diagnosis and treatment method information t+1, and the predicted patient disease information t+1 to the predicted patient management policy t+N, the predicted diagnosis and treatment method information t+N, and the predicted patient disease information t+N for the diagnosis and treatment plan patterns 1 to M (M is an integer greater than 2).

[0210] Treatment plan patterns 1 through M are patterns derived from extracting possible treatment approaches for the target patient. These patterns include not only treatment approaches predicted based on the target patient's characteristics, but also treatment plan patterns used for patients with similar characteristics. This is done to predict the therapeutic effects (disease information) of potential treatment plan patterns, enabling doctors to select the optimal treatment option.

[0211] <Details of the decision (forecast) on inspection items, etc.>

[0212] Figure 16 This flowchart illustrates the processing of multiple patterns of patient management policies, treatment methods, and patient disease information from the second embodiment, estimating the time from time point t+1 (time step t+1) to time point t+N (time step t+N). In the following description, the treatment prediction unit 1500 is the main actor in each step. However, the treatment planning unit 1100, which includes the treatment prediction unit 1500, is implemented by expanding the program in the processor 101. Therefore, the processor 101 may also be the main actor.

[0213] (i)S1601

[0214] The diagnosis and treatment prediction unit 1500 obtains (accepts) patient information (patient information of the patient being treated) and patient management policies from time steps tk to t from the medical information server 111. The patient information from time steps tk to t includes patient attribute information, diagnosis and treatment method information from time steps tk to t, examination result information from time steps tk to t, and patient disease information from time steps tk to t. Furthermore, the diagnosis and treatment prediction unit 1500 sets the learned model 1303 (preparing the learned model 1303 in a manner that enables its use).

[0215] (ii)S1602

[0216] The diagnosis and treatment prediction unit 1500 integrates the patient information obtained in S1601 and extracts the characteristic values ​​of the patient information. Specifically, the diagnosis and treatment prediction unit 1500 converts the received patient information into text and digitizes it in a manner that can be processed by a neural network (integration of the patient information), applies the integrated patient information to the neural network (e.g., deep learning), and thereby calculates the characteristic values ​​of the numerical information.

[0217] (iii)S1603

[0218] The treatment prediction unit 1500 applies the feature value of the patient information obtained in S1602 to the learned model 1303, thereby inferring the patient management policy, patient disease information, and treatment method from time step t+1 to time step t+N as a treatment plan.

[0219] Regarding the treatment plan, a plurality of patterns (pattern 1 to pattern M) are estimated in each of the time steps t+1 to t+N.

[0220] (iv)S1604

[0221] The output device 106 receives the information estimated in S1603 (patient management policies and treatment methods for each patient's disease information from time step t+1 to time step t+N) from the diagnosis and treatment prediction unit 1500, and sends the information as a diagnosis and treatment plan (mode 1 to mode M) to the information terminal 110.

[0222] The information terminal 110 displays the received treatment plan information on a display screen, for example, so that the user (doctor) can select a desired treatment plan.

[0223] <Data Flow of Learning Processing by the Inspection Learning Unit 1700 of the Inspection Planning Unit 1012>

[0224] Figure 17This is a diagram for explaining the data flow in the learning process of the inspection learning unit 1700 of the inspection planning unit 1012 of the second embodiment. The inspection planning unit 1012 includes: the inspection learning unit 1700, which generates a learning completion model; and the inspection prediction unit 400, which uses the learning completion model to infer inspection items, etc. Figure 17 In the example, the inspection learning unit 1700 generates various data and outputs a learning model 1701 generated based on these data.

[0225] When generating a learning completion model, the inspection learning unit 1700 obtains as input data the criteria information, the patient management policy at time point t (time step t), the patient management policy t+N at time point t to time point t+N (time step t+N), patient attribute information, inspection items, and inspection result information t+N at time point t (time step t), inspection result information t+N at time point t+1 (time step t+1), patient disease information t+1 to time point t+N (time step t+N), and treatment method t+N at time point t+1 (time step t+1).

[0226] The inspection learning unit 1700 uses a probability distribution model to collect inspection result data for each inspection item and estimates the probability distribution of the inspection results 1702 (estimation of probability distribution of inspection results). As the number of data increases, the probability distribution approaches a normal distribution.

[0227] In addition, the inspection learning unit 1700 generates a probability distribution function through learning based on the estimated probability distribution 1702, and uses the probability distribution function to generate the restored inspection results 1703_1 at the t+1 time point (time step t+1) to the restored inspection results 1703_N at the t+N time point (time step t+N), the inspection value 1704_1 of each inspection item at the t+1 time point (time step t+1) to the t+N time point (time step t+N) 1704_N, and the disease diagnosis probability and risk 1705_1 at the t+1 time point (time step t+1) to the t+N time point (time step t+N) The probability and risk of disease diagnosis 1705_N. Based on this information, the inspection and learning unit 1700 estimates patient management policies 1706_1 at time point t+1 (time step t+1) to patient management policies 1706_N at time point t+N (time step t+N), patient disease information 1707_1 at time point t+1 (time step t+1) to patient disease information 1707_N at time point t+N (time step t+N), and treatment methods 1708_1 at time point t+1 (time step t+1) to treatment methods 1708_N at time point t+N (time step t+N). Furthermore, the inspection and learning unit 1700 compares the estimated patient management policies, patient disease information, and treatment methods with the actual patient management policies, patient disease information, and treatment methods, and estimates a learning error 1709_N at time point t+N (time step t+N) based on the learning error 1709_1 at time point t (time step t).

[0228] When the learning error t+N converges from the learning error t, the inspection learning unit 1700 outputs a learning completion model 1701 having parameters related to the converged learning error.

[0229] <Details of Learning Parameter Determination Process>

[0230] Figure 18 This is a flowchart for describing the details of the learning parameter determination process of the second embodiment. In the following description, the inspection learning unit 300 is the main body of each step. However, since the inspection planning unit 1012 including the inspection learning unit 1700 is implemented by developing a program in the processor 101, the processor 101 may also be the main body of the operation.

[0231] (i)S1801

[0232] The examination learning unit 1700 receives patient attribute information, examination item information, and examination result information t at time point t (time step t) required for generating the learned model 1701 .

[0233] (ii)S1802

[0234] The examination learning unit 1700 integrates the information obtained in S1801 and extracts the characteristic values ​​of the information. Specifically, the examination learning unit 1700 converts the received patient attribute information, examination item information, and examination result information t at time point t (time step t) into text, converts the text into numerical values ​​in a manner that can be processed by a neural network (information integration), and applies the integrated information to the neural network (e.g., deep learning) to calculate the characteristic values ​​of the numerical information.

[0235] (iii)S1803

[0236] The inspection learning unit 1700 collects a plurality of inspection results for each inspection item and estimates a probability distribution (function) for each inspection item using a probability distribution model.

[0237] (iv)S1804

[0238] The inspection learning unit 1700 restores the inspection result of the next time step (t→t+1) based on the probability distribution estimated in S1703. This is because the inspection result as the correct value is known, so it is a process to confirm whether the correct value is obtained based on the estimated probability distribution.

[0239] (v)S1805

[0240] The examination learning unit 1700 estimates the examination value, disease diagnosis probability, and risk based on the examination results restored in S1804. If the original examination results can be restored based on the estimated probability distribution, the estimated probability distribution is accurately estimated. In S1805, the examination value, disease diagnosis probability, and risk are estimated to what extent they would be if new patient information were input.

[0241] (vi)S1806

[0242] The inspection learning unit 1700 uses the restored inspection result estimated in S1804 to estimate the learning error of the inspection result at the next time step (t→t+1). Here, the learning error refers to the error between the original inspection result and the restored inspection result.

[0243] (vii)S1807

[0244] The inspection learning unit 1700 applies the inspection value, disease diagnosis probability and risk estimated in S1805 to, for example, a deep learning method (RNN) for time series learning, thereby estimating the patient management policy, disease information and treatment method for the next time step (t→t+1).

[0245] (viii)S1808

[0246] The inspection learning unit 1700 estimates a learning error between the actual treatment method and the treatment method for the next time step (time step t+1) estimated in S1807.

[0247] (ix)S1809

[0248] The inspection learning unit 1700 determines whether the processing is complete for all time steps (t to t+N). If the processing is complete (if yes in S1809), the process moves to S1810. If there is information about time steps that have not been processed (if no in S1809), the process moves to S1811.

[0249] (x)S1810

[0250] The learning unit 1700 checks whether the learning error estimated in S1806 and the learning error estimated in S1808 have converged. Convergence can be determined, for example, by determining whether the learning error is less than a preset threshold. If the learning error value has converged (if YES in S1810), the process proceeds to S1813. If the learning error value has not converged (if NO in S1810), the process proceeds to S1814.

[0251] (xi)S1811

[0252] The inspection learning unit 1700 updates the current time step (for example, t) to the next time step (for example, t+1).

[0253] (xii)S1812

[0254] The inspection learning unit 1700 obtains the inspection items and inspection result information of the current time step + 1 (eg, t + 1). The process then moves to S1802, where information integration and feature extraction are performed again.

[0255] (xiii)S1813

[0256] The inspection learning unit 1700 sets the learning parameters for which the learning error converges to the learning completed model. Here, the learning parameters include neural network parameters within the RNN for predicting patient management policies, treatment information, and disease information.

[0257] (xiv)S1814

[0258] The inspection learning unit 1700 updates the learning parameters (for example, parameters of the neural network) according to the optimization function for updating so as to reduce the learning error, and executes the process from S1801 again.

[0259] <Output screen for treatment plan and examination plan>

[0260] Figure 19 1 is a diagram showing a configuration example of an output screen (UI: User Interface) 1900 for a treatment plan and an examination plan according to the second embodiment. This output screen 1900 can be displayed on the display screen of the information terminal 110, for example.

[0261] Output screen 1900 is a screen that displays the recommended contents of the diagnosis and treatment plan and the examination plan. Specifically, it is a screen that displays the examination items (estimated) and treatment methods (estimated) for the process from "suspected" to "completely cured" related to a specific disease. Output screen 1900 is composed of, for example, an initial diagnosis information display column 1901, a follow-up diagnosis information display column 1902, a process observation information display column I_1903, a process observation information display column II_1904, a patient information update button 1905, a plan update button 1906, and an examination instruction output button 1907.

[0262] The patient information update button 1905 is a button that is pressed when inputting patient information of a target patient when the target patient is changed, for example.

[0263] The plan update button 1906 is a button that is pressed when the process of creating (predicting) a treatment plan is executed again.

[0264] The inspection instruction output button 1907 is a button that is pressed when an inspection instruction is issued to an inspection room (not shown) via the inspection sequencing system 1110 .

[0265] Figure 19 The output screen 1900 shows the following content: when disease A is suspected in the initial diagnosis, the diagnosis and treatment plan and the examination plan are predicted, (i) the disease A is confirmed through follow-up visits, and examinations 4, 5 and 6 are recommended, and then prescriptions a, b and c are recommended as treatment methods; (ii) during treatment, examinations 3, 4 and 5 are recommended, and then a, b and c are recommended as treatment methods; (iii) complete cure is achieved by continuing the examinations and treatments represented by process observation I.

[0266] <Configuration Example of GUI Displayed on the Display Screen of Information Terminal 110>

[0267] Figure 20 This figure shows a configuration example of a test order GUI 2000 displayed on the display screen of the information terminal 110 according to the second embodiment. Through the test order GUI 2000, patient management policies are selected and displayed (doctor selection), and disease determination results and test items are output.

[0268] The examination order GUI 2000 includes, for example, a management policy selection / display unit 1001, a disease probability display unit 2002, and a disease examination item display unit 2003. The difference from the examination order GUI 1000 of the first embodiment is that the disease examination item display unit 2003 of the examination order GUI 2000 is configured to enable comparison of patient costs and hospital burdens between the current period and one month later (in the future).

[0269] When receiving recommended examination information from the computer 100 of the patient care plan determination system 10 , the management policy selection / display unit 2001 displays the patient management policy of the target patient that can be selected by the doctor and the currently selected patient management policy.

[0270] The disease probability display unit 2002 provides information on the disease name, disease probability, and disease risk (how high is the risk of the disease itself) obtained through examinations and disease prediction processing (prediction simulation). For example, if the prediction results indicate that disease A is likely to occur with a 70% probability, the risk of the disease is indicated as medium.

[0271] The disease examination item display unit 2003 displays, for example, examination items for diseases with a disease probability higher than a predetermined value or examination items for high-risk diseases, patient costs indicating the fees paid by the patient for the examination, and hospital burden indicating the cost burden on the hospital side. Figure 10 In this example, the test items for disease A, which has a 70% probability, and disease C, which has a high risk, are displayed. For disease A, which has the highest probability, the patient cost remains the same both now and one month from now, and the hospital burden is zero one month from now. In contrast, for disease C, which has a low probability but the highest risk, the patient cost remains the same both now and one month from now, and the hospital burden is zero.

[0272] (3) Summary

[0273] (i) The patient care plan determination system 10 of this embodiment is a system that uses a learning completion model to estimate future examination item candidates and their values ​​(examination values) based on patient management policies, patient information, and information on examination items and examination results. Figure 1 and Figure 2As shown, the patient care plan determination system 10 can be composed of a computer 100, an information terminal 110, and a medical information server (external recording device) 111. The computer 100 performs the following processing: obtaining specified information from the medical information server and the information terminal, and determining candidate examination items related to the target patient's disease. Specifically, the computer applies the target patient's patient attribute information, the target patient's examination results, and the target patient's patient management policy to a learning model (examination plan learning model) 310 for calculating the examination value of each examination item, thereby estimating the examination value of the target patient's future examination item candidates and outputting the target patient's examination item candidates and the examination value, wherein the examination value represents information that can identify the disease according to the patient management policy. The output destination can be, for example, the information terminal 110 operated by a user (doctor, etc.). In addition, the examination item with the highest examination value can be selected as the candidate examination item. In this way, it is possible to recommend highly appropriate examinations based on the results of each timing of examination, diagnosis, treatment, and process observation in the overall patient management workflow.

[0274] (ii) The learned model 310 includes parameters for estimating the probability distribution of the test results for each test item, parameters for reconstructing the test results based on the estimated probability distribution, and parameters for estimating the test value as learning parameters. The process of generating the learning parameters is outlined as follows. Specifically, the computer 100 integrates patient information from multiple patients (learning sample data) to extract features and estimates the probability distribution of the test results for each test item. The computer 100 then reconstructs the test results based on the estimated probability distribution to generate reconstructed test results, and estimates the test value based on the results. Furthermore, the computer 100 estimates the learning error, which is the error between the test value obtained from the actual test results and the estimated test value. If the learning error converges, the learning parameter that provides the learning error is set as the learning parameter of the learned model 310. This allows the construction of a learned model 310 that uses the information of the target patient (patient attribute information, the target patient's test results, and the target patient's patient management policy) to calculate future test item candidates and information indicating their appropriateness. Furthermore, the convergence of the learning error can be determined using a threshold. When making this judgment, an appropriate threshold must be used. Therefore, computer 100 uses information about the diagnosis probability of a predetermined disease and the risk of that disease, estimated based on the recovery test results, to estimate a patient management policy. Based on the comparison between this estimated patient management policy and the actual management policy, it estimates a threshold for determining whether the learning error has converged. This allows the threshold used in determining whether the learning error has converged to be set to an appropriate value.

[0275] (iii) The computer 100 estimates a medical remuneration score based on the candidate examination items, the patient management policy of the target patient, and the medical remuneration calculation conditions obtained from the medical information server. Based on the medical remuneration score, the hospital's burden of cost is estimated. This allows for cost-effectiveness considerations when deciding future examination items.

[0276] (iv) In addition to the examination plan, the second embodiment also describes a patient care plan determination system 10 that provides a treatment plan. Specifically, the computer 100 obtains patient management policies for the target patient in the past (time step tk) and at the current time point (time step t), treatment method information for the target patient in the past and at the current time point, patient attribute information for the target patient, examination items and examination result information for the target patient in the past and at the current time point, and patient disease information for the target patient in the past and at the current time point from the medical information server 111, and applies the obtained information to a learning completion model (treatment plan learning completion model) 1303 for predicting the treatment plan for the target patient based on the patient examination and treatment history information of the target patient, thereby inferring a treatment plan that includes the patient management policies and treatment methods for the target patient in the future (time step t+1 to time step t+N). Thus, based on appropriate examinations, the doctor can also propose appropriate treatment methods for the patient at each future time point (even if there is a lack of experience). Furthermore, the computer 100 performs multiple model estimations of the treatment plan at each future time point (time step t+1 to time step t+N) of the target patient (see Figure 15 Then, the information terminal 110 displays the information of the treatment plan of the target patient received from the computer 100 in a time series on the display screen. For example, the information of the treatment plan is displayed on the display screen in the order of initial visit, follow-up visit and process observation.

[0277] (v) The outline of the process for generating learning parameters for the learned model 1303 is as follows. Specifically, the computer 100 obtains from the medical information server 111 patient management policies for multiple patients (samples for learning model generation) prior to a predetermined time point (time step tk to time step t), diagnosis and treatment method information for multiple patients prior to a predetermined time point, patient attribute information for multiple patients, examination items and examination result information for multiple patients prior to a predetermined time point, and patient disease information for multiple patients prior to a predetermined time point as patient examination and treatment history information for learning model generation, extracts feature quantities from this information, performs machine learning based on this feature quantity, and estimates the patient management policies, diagnosis and treatment methods, and patient disease information for multiple patients after the predetermined time point (time step t+1 to time step t+N). In addition, the computer 100 compares the actual patient management policies, diagnosis and treatment methods, and patient disease information of multiple patients after a specified time point with the estimated patient management policies, diagnosis and treatment methods, and patient disease information after a specified time point, thereby estimating a learning error for the diagnosis and treatment plan. When the learning error converges, the learning parameter providing the learning error for the diagnosis and treatment plan is set as the learning parameter of the learning completion model 1303. In addition, when the learning error does not converge, the computer 100 performs the following processing: using the optimization function to update the learning parameter providing the learning error for the diagnosis and treatment plan, and using the updated learning parameter to estimate again. Thus, it is possible to construct a learning completion model 1303 for appropriately proposing future diagnosis and treatment plans using the information of the patient to whom the diagnosis and treatment plan is applied (patient attribute information, test results of the target patient, patient management policy of the target patient).

[0278] (vi) The functions of the present embodiment and each embodiment can also be implemented by the program code of software. In this case, a storage medium recording the program code is provided to a system or device, and the computer (or CPU, MPU) of the system or device reads the program code stored in the storage medium. In this case, the program code read from the storage medium itself implements the functions of the embodiment described, and the program code itself and the storage medium storing the program code constitute the present disclosure. As a storage medium for supplying such a program code, for example, a floppy disk, CD-ROM, DVD-ROM, hard disk, optical disk, optical magnetic disk, CD-R, magnetic tape, non-volatile memory card, ROM, etc. are used.

[0279] Alternatively, an OS (operating system) or the like running on a computer may perform part or all of the actual processing according to the instructions of the program code, thereby realizing the functions of the embodiments described above through such processing. Alternatively, after the program code is read from a storage medium and written to a memory on a computer, the CPU or the like of the computer may perform part or all of the actual processing according to the instructions of the program code, thereby realizing the functions of the embodiments described above through such processing.

[0280] Furthermore, the program code of the software that implements the functions of the implementation methods and each embodiment can be distributed via a network and stored in a storage unit such as a hard disk or memory of the system or device or a storage medium such as a CD-RW or CD-R. When in use, the computer (or CPU, MPU) of the system or device reads and executes the program code stored in the storage unit or the storage medium.

[0281] The processes and techniques described herein are not inherently associated with any specific device and can also be implemented by combining various components. In addition, various types of general-purpose devices can also be added. In order to perform the functions of this embodiment and each example, a dedicated device can also be constructed. In addition, various functions can also be formed by appropriately combining multiple structural elements disclosed in this embodiment and each example. For example, several structural elements can be deleted from all the structural elements shown in the embodiment and each example, and the structural elements of different examples can also be appropriately combined.

[0282] While specific embodiments are described in this disclosure, these are for illustrative purposes only (for understanding the techniques of this disclosure) and are not intended to be limiting. A person with ordinary skill in the art will recognize that there are many combinations of hardware, software, and firmware suitable for implementing the techniques of this disclosure. For example, the described software can be implemented using a wide range of programming or scripting languages, such as assembler, C / C++, Perl, Shell, PHP, and Java (registered trademark).

[0283] Furthermore, in the above-described embodiments, the control lines and information lines are those considered necessary for explanation, and not all control lines and information lines are necessarily shown in the product. All components may be connected to each other.

[0284] Furthermore, anyone with ordinary knowledge in this technical field will be able to clearly understand other implementations of the present disclosure by examining the present embodiment and each example. The description and specific examples are merely typical examples, and the technical scope and spirit of the present disclosure are expressed in the following claims.

[0285] Captions

[0286] 10 Patient Care Plan Decision System

[0287] 100 Computer

[0288] 101 Processor

[0289] 102 main storage device

[0290] 103 Secondary storage device

[0291] 104 Network Adapter

[0292] 105 Input Device

[0293] 106 Output Device

[0294] 109 Network

[0295] 110 Information Terminal

[0296] 111 external recording device, medical information server

[0297] 1011 Patient Information Acquisition Department

[0298] 1012 Inspection and Planning Department

[0299] 1101 Check command output unit

[0300] 1100 Treatment Planning Department

[0301] 1110 Check sorting system.

Claims

1. A patient care plan determination system for determining at least candidate examination items related to a disease of a target patient, characterized in that: The patient care plan decision system has: a medical information server that stores at least attribute information and examination results of a plurality of patients; an information terminal that provides a patient management policy indicating a classification of the target patient; a computer that acquires predetermined information from the medical information server and the information terminal and determines candidate examination items related to the disease of the target patient; The computer executes: a process of acquiring attribute information of the target patient and a patient management policy of the target patient from the medical information server; applying patient attribute information of the subject patient, the examination result of the subject patient, and the patient management policy of the subject patient to an examination plan learning completion model for calculating the examination value of each examination item, thereby estimating the examination value of a future candidate examination item for the subject patient, the examination value indicating information capable of identifying a disease according to the patient management policy; as well as A process of outputting the examination item candidates and the examination value of the target patient.

2. The patient care plan determination system according to claim 1, wherein: The computer sets the examination item with the greatest examination value as the examination item candidate for the target patient.

3. The patient care plan determination system according to claim 1, wherein: The inspection plan learning completion model includes, as learning parameters, parameters for estimating a probability distribution in the inspection result of each inspection item, parameters for restoring the inspection result based on the estimated probability distribution, and parameters for estimating the inspection value.

4. The patient care plan determination system according to claim 3, wherein: said computer, a process of acquiring patient information of a plurality of patients from the medical information server, the patient information including attribute information of the patients, the examination results, and the patient management policy; integrating patient information of the plurality of patients and extracting feature quantities; a process of estimating a probability distribution of the examination results of the plurality of patients according to the examination items; Reconstructing the inspection result based on the estimated probability distribution to generate a reconstructed inspection result; A process of estimating the inspection value based on the restored inspection result; Processing of estimating a learning error, wherein the learning error is the difference between the inspection value obtained based on the actual inspection result and the estimated inspection value; When the learning error converges, a learning parameter of the learning error is provided as a learning parameter of the inspection plan learning completion model.

5. The patient care plan determination system according to claim 4, wherein: The computer estimates a patient management policy using information on the diagnosis probability of a predetermined disease and the risk of the disease estimated based on the restoration test results, and estimates a threshold for determining whether the learning error has converged based on a comparison result between the estimated patient management policy and the actual management policy.

6. The patient care plan determination system according to claim 1, wherein: The computer also executes: A process of acquiring receipt information including information on medical treatment remuneration calculation conditions from the medical information server; a process of estimating a medical treatment remuneration score based on the candidate examination items, the patient management policy of the target patient, and the medical treatment remuneration calculation conditions; as well as The process of estimating the cost borne by the hospital based on the medical treatment remuneration score.

7. The patient care plan determination system according to claim 1, wherein: The computer also executes: Acquiring patient examination and treatment history information from the medical information server, the patient examination and treatment history information including patient management policies of the subject patient at past and current time points, diagnosis and treatment method information of the subject patient at past and current time points, patient attribute information of the subject patient, examination items and examination result information of the subject patient at past and current time points, and patient disease information of the subject patient at past and current time points; a process of inferring a treatment plan including a future patient management policy and treatment methods for the subject patient by applying the subject patient's past and current examination and treatment history information to a treatment plan learning completion model, the treatment plan learning completion model being used to predict the subject patient's treatment plan based on the subject patient's examination and treatment history information; as well as A process of outputting a treatment plan for the target patient.

8. The patient care plan determination system according to claim 7, wherein: The computer further executes: a process of extracting a feature value of the patient's examination and medical treatment history information; The feature amount of the patient examination and treatment history information is applied to the treatment plan learning completion model.

9. The patient care plan determination system according to claim 7, wherein: The computer estimates the treatment plan as a plurality of patterns at each future time point of the target patient.

10. The patient care plan determination system according to claim 7, wherein: The information terminal receives information on the treatment plan of the target patient from the computer, and displays the contents of the treatment plan on a display screen in a time series manner.

11. The patient care plan determination system according to claim 10, wherein: The information terminal displays the contents of the diagnosis and treatment plan on the display screen in the order of initial diagnosis, follow-up diagnosis, and process observation.

12. The patient care plan determination system according to claim 7, wherein: The computer generates the treatment plan learning completion model by executing the following processing: Obtaining from the medical information server the patient management policies of a plurality of patients before a specified time point, the diagnosis and treatment method information of the plurality of patients before the specified time point, the patient attribute information of the plurality of patients, the examination items and examination result information of the plurality of patients before the specified time point, and the patient disease information of the plurality of patients before the specified time point as processing of the patient examination and diagnosis and treatment history information for learning completion model generation; Processing of extracting feature quantities for generating a learning model by integrating the patient examination and medical history information for generating a learning model; Performing machine learning based on the feature quantity for generating the learning completion model to estimate the patient management policy, diagnosis and treatment method, and patient disease information processing after the predetermined time point; Inferring a handling of learning errors with respect to a treatment plan by comparing actual patient management policies, treatment methods, and patient disease information of the plurality of patients after the predetermined time point with the estimated patient management policies, treatment methods, and patient disease information after the predetermined time point; as well as A process of setting a learning parameter providing the learning error for the treatment plan as a learning parameter of the treatment plan learning completion model when the learning error for the treatment plan converges.

13. The patient care plan determination system according to claim 12, wherein: The computer executes a process of updating a learning parameter that provides the learning error for the treatment plan using an optimization function when the learning error for the treatment plan has not converged, and performing the estimation again using the updated learning parameter.

14. A method for determining a patient care plan, comprising determining at least candidate examination items related to a disease of a target patient, wherein: The patient care plan determination method comprises: A computer that obtains predetermined information from a medical information server that stores at least attribute information and examination results of a plurality of patients and an information terminal that provides a patient management policy indicating the classification of the target patient, and determines candidate examination items related to the disease of the target patient, and obtains the attribute information of the target patient and the patient management policy of the target patient from the medical information server; the computer estimates the examination value of a future candidate examination item for the subject patient by applying the subject patient's patient attribute information, the subject patient's examination result, and the subject patient's patient management policy to an examination plan learning completion model for calculating the examination value of each examination item, the examination value indicating information capable of identifying a disease according to the patient management policy; The computer outputs examination item candidates for the target patient and the examination value.

Images