Systems, programs, and methods for predicting clinical outcomes of diseases

Abstract

To provide a technique capable of accurately predicting changes in physical conditions, including clinical outcomes.SOLUTION: The system for predicting clinical outcome of disease includes an information processing device that executes: a first program that causes an information processing device to execute a first information processing method of predicting the subsequent clinical outcome of a patient to be predicted from the examination data at the first visit; and a second program that causes the information processing device to execute a second information processing method of predicting the subsequent clinical outcome of the patient to be predicted from the examination data during hospitalization.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a system, a program, and a method for predicting the clinical outcome of a disease.

Background Art

[0002] The populations of 25 countries around the world have started to decline. It is predicted that the total global population will head towards decline by the mid-21st century. This means that the aging population and low birth rate are progressing on a global scale. Due to the tax burden used for the medical care and pensions of a large number of elderly people, the lives of the young are becoming difficult. When trying to solve the problem of illness with high-cost advanced medical care, the national finances become strained, creating a situation where only economically advantaged people can receive treatment. In order to realize sustainable medicine where all people can receive appropriate medical and welfare services, it is essential to develop technologies for individually predicting and preventing the onset of diseases (Non-Patent Document 1).

[0003] Since the occurrence of the acute respiratory disease COVID-19 caused by the SARS-CoV-2 virus in December 2019, the infection has been spreading continuously. As of April 2021, more than 150 million infected people and more than 3 million deaths have been confirmed worldwide (Non-Patent Document 2). COVID-19 is characterized by a wide range of symptoms, where most infected people end up asymptomatic or with mild symptoms such as cough and fever, while some affected individuals rapidly develop pneumonia and require mechanical ventilation (Non-Patent Document 3). With the continuous influx of COVID-19 patients, the shortage of medical resources in medical institutions around the world has become severe, and the optimal allocation of medical devices such as ventilators and ECMO (extracorporeal membrane oxygenation), drugs, and staff has become an urgent task.

[0004] Since the severity of COVID-19 is closely related to the allocation of medical resources and staff, predicting the severity prognosis and the timing of severe illness would allow for the preparation of optimal medical resources in advance. Numerous prognostic scores and prognostic prediction models have been proposed regarding the severity and mortality of COVID-19 (Non-Patent Literature 3). However, all of these prediction models predict the prognosis using data at the time of admission and do not take into account changes in the patient over time, thus limiting their predictive accuracy. [Prior art documents] [Non-patent literature]

[0005] [Non-Patent Document 1] Stein Emil Vollset, Emily Goren, Chun-Wei Yuan, Jackie Cao, Amanda E Smith, Thomas Hsiao, Catherine Bisignano, Gulrez S Azhar, Emma Castro, Julian Chalek, Andrew J Dolgert, Tahvi Frank, Kai Fukutaki, Simon I Hay, Rafael Lozano, Ali H Mokdad, Vishnu Nandakumar, Maxwell Pierce, Martin Pletcher, Toshana Robalik, Krista M Steuben, Han Yong Wunrow, Bianca S Zlavog, Christopher JL Murray (2020) Fertility, mortality, migration, and population scenarios for 195 countries and territories from 2017 to 2100:a forecasting analysis for the Global Burden of Disease Study Lancet 2020; 396: 1285-306 [Non-Patent Document 2] Dong E, Du H, Gardner L. An interactive web-based dashboard to track COVID-19 in real time. Lancet Infect Dis 2020;20(5):533-4. [Non-Patent Document 3] Menni C, Valdes AM, Freidin MB, et al. Real-time tracking of self-reported symptoms to predict potential COVID-19. Nat Med 2020;26(7):1037-40. [Non-Patent Document 4] Wynants L, Van Calster B, Collins GS, et al. Prediction models for diagnosis and prognosis of covid-19 infection: systematic review and critical appraisal. BMJ 2020;369:m1328. [Overview of the Initiative] [Problems that the invention aims to solve]

[0006] The rapid increase in age-related diseases, including dementia, due to the aging of society, the migration of wild animals towards the poles due to rapid global warming, and the migration of wild animals due to deforestation are disrupting the zoning between wild animals and humans / livestock, and it is thought that emerging infectious diseases, such as COVID-19, will become more frequent in the future. In the midst of this rapidly changing disease structure, achieving sustainable healthcare requires expanding intervention methods and timing from the current approach of treating illnesses after they have developed to detecting early signs and implementing preventative measures. This necessitates highly accurate predictive technologies that can identify signs that people might miss in conventional clinical processes and provide insights for intervention.

[0007] Predicting the progression of macroscopic states of the entire body is essential for highly accurate, personalized prevention and proactive medicine based on predictions. However, there are no existing principles or methods for predicting such macroscopic state progressions. The knowledge system of biomedical science consists of microscopic molecular models based on causal relationships. Macroscopic state transitions based on this knowledge are explained using simulations and network models. However, this explanation approximates biological systems, which are nonlinear, non-equilibrium open systems, with linear models. Many existing scores and models are based on statistical models such as linear and additive models, and do not take into account the heterogeneity of individual patients or the interactions between explanatory variables, making them unsuitable for highly accurate predictions.

[0008] The present invention aims to provide a technology for accurately predicting changes in physical condition, including clinical outcomes, based on previously unnoticed signs. [Means for solving the problem]

[0009] To solve the aforementioned problems, the inventors considered that a method for understanding the macroscopic states that a person can take is effective as a prerequisite for predicting physical conditions, including clinical outcomes. This list of possible states can be used to predict individual state transitions through inductive reasoning. In this case, it is effective to determine the state transition probability. Information geometry has shown that in nonlinear systems where components interact, the possible states are not based on prerequisites of the state space but rather on geometric constraints. To understand the constraints on possible states, it is effective to apply this concept to state identification. This requires an identification method that replaces different states with distances. The decision tree method is suitable for this purpose because it can determine distance by the number of branches traversed, and it can also identify nonlinear interactions. On the other hand, when predicting changes in physical conditions, including clinical outcomes, with high accuracy in a clinical setting, it is necessary to dynamically predict by combining state identification at the initial consultation with state identification based on time-series data at hospitalization or follow-up consultations. Therefore, the first step in solving this problem is to create treatment and prevention plans based on state identification at the initial consultation. In this case, we considered methods such as unsupervised random forests and gradient boosting decision trees to be effective, and by using gradient boosting decision trees, a type of ensemble machine learning, we developed a COVID-19 severity prognosis prediction model based on examination data obtained at the initial consultation. We then found that we could predict the subsequent clinical outcome of the target patients using the examination data from the initial consultation of COVID-19 patients and this severity prognosis prediction model.

[0010] Next, by observing the temporal progression of the patient's physical condition from the initial consultation, it is necessary to predict changes in physical condition, including clinical outcomes, with high accuracy. This prediction is fundamentally different from the prediction at the initial consultation because interventions, such as treatment, are carried out during the observation period. We believed that the Random Survival Forest method would be effective in such predictions, and developed a dynamic mortality risk model by performing machine learning survival analysis based on the temporal clinical data of hospitalized patients. We then found that by using the examination data of hospitalized patients at multiple time points and this dynamic mortality risk model, we could predict the risk of death from COVID-19 in the target patients with high accuracy. These new findings demonstrated that state identification using distance information with decision trees is effective in predicting changes in physical condition, including clinical outcomes, with high accuracy in various clinical situations. Since this method is applicable to any disease, we were able to complete the present invention.

[0011] To solve the aforementioned problems, one aspect of the present invention relates to a method, system, or program for obtaining highly accurate clinical outcomes from patient data, characterized by incorporating two axes—a prediction axis of state transitions and a variation axis of actual state transitions—into a discrete bodily state transition scheme.

[0012] To solve the aforementioned problems, a clinical outcome prediction system for a disease according to one aspect of the present invention includes a first program that causes an information processing device to execute a first information processing method that predicts the subsequent clinical outcome of a target patient from examination data at the initial consultation, A second program causes an information processing device to execute a second information processing method that predicts the subsequent clinical outcome of a target patient based on test data obtained during hospitalization. It includes an information processing device that performs the following: The first information processing method is A first training data acquisition step involves obtaining training data that includes one or more pairs of initial examination data and clinical outcomes of patients suffering from the target disease, A first learning step involves using the initial examination data as input to train a first predictive model that predicts the clinical outcome, by referring to the training data. A first test data acquisition step involves acquiring test data from the patient to be predicted at the time of their first visit, A first prediction step involves inputting the test data acquired in the first test data acquisition step into the first prediction model to predict the subsequent clinical outcome of the patient to be predicted, Includes, The second information processing method described above is: A second training data acquisition step involves obtaining training data that includes one or more pairs of laboratory data from patients suffering from the target disease during hospitalization and their clinical outcomes. A second learning step involves using test data from multiple points in time during hospitalization as input to train a second predictive model that predicts the clinical outcome, by referring to the training data. A second test data acquisition step involves acquiring test data at multiple points in time during the hospitalization of the patient to be predicted, A second prediction step involves inputting the test data acquired in the second test data acquisition step into the second prediction model to predict the subsequent clinical outcome of the patient being predicted, It is a composition that includes [this].

[0013] To solve the aforementioned problems, a clinical outcome prediction system for a disease according to one aspect of the present invention includes a first program that causes an information processing device to execute a first information processing method that predicts the subsequent clinical outcome of a target patient from examination data at the initial consultation, A second program causes an information processing device to execute a second information processing method that predicts the subsequent clinical outcome of a target patient based on test data obtained during hospitalization. It includes an information processing device that performs the following: The first information processing method is A first test data acquisition step involves acquiring test data from the patient to be predicted at the time of their first visit, Using the examination data at the first visit as input, for the first prediction model that predicts the clinical outcome, by inputting the examination data obtained in the examination data acquisition step, a first prediction step of predicting the subsequent clinical outcome of the patient to be predicted; including The second information processing method A second examination data acquisition step of acquiring examination data at a plurality of points in time during the hospitalization of the patient to be predicted; Using the examination data at a plurality of points in time during hospitalization as input, for the second prediction model that predicts the clinical outcome, by inputting the examination data obtained in the examination data acquisition step, a second prediction step of predicting the subsequent clinical outcome of the patient to be predicted; It is a configuration including

[0014] To solve the above problems, a clinical outcome prediction program for a disease according to an aspect of the present invention includes a first program that causes an information processing device to execute a first information processing method for predicting the subsequent clinical outcome of a patient to be predicted from examination data at the first visit, A second program that causes an information processing device to execute a second information processing method for predicting the subsequent clinical outcome of a patient to be predicted from examination data during hospitalization, including The first information processing method A first teacher data acquisition step of acquiring teacher data including one or more pairs of examination data at the first visit of a patient suffering from the target disease and the clinical outcome; A first learning step of using the examination data at the first visit as input and training a first prediction model for predicting the clinical outcome by referring to the teacher data; A first examination data acquisition step of acquiring examination data at the first visit of the patient to be predicted; A first prediction step of predicting the subsequent clinical outcome of the patient to be predicted by inputting the examination data obtained in the first examination data acquisition step to the first prediction model; including The second information processing method A second training data acquisition step involves obtaining training data that includes one or more pairs of laboratory data from patients suffering from the target disease during hospitalization and their clinical outcomes. A second learning step involves using test data from multiple points in time during hospitalization as input to train a second predictive model that predicts the clinical outcome, by referring to the training data. A second test data acquisition step involves acquiring test data at multiple points in time during the hospitalization of the patient to be predicted, A second prediction step involves inputting the test data acquired in the second test data acquisition step into the second prediction model to predict the subsequent clinical outcome of the patient being predicted, It is a configuration that includes [this].

[0015] To solve the aforementioned problems, a clinical outcome prediction program for a disease according to one aspect of the present invention includes a first program that causes an information processing device to execute a first information processing method that predicts the subsequent clinical outcome of a target patient from examination data at the initial consultation, A second program causes an information processing device to execute a second information processing method that predicts the subsequent clinical outcome of a target patient based on test data obtained during hospitalization. Includes, The first information processing method is A first test data acquisition step involves acquiring test data from the patient to be predicted at the time of their first visit, A first prediction step predicts the subsequent clinical outcome of the patient being predicted by inputting the examination data acquired in the examination data acquisition step into a first prediction model that uses the examination data from the initial examination as input to predict the clinical outcome, Includes, The second information processing method described above is: A second test data acquisition step involves acquiring test data at multiple points in time during the hospitalization of the patient to be predicted, A second prediction step involves inputting the test data acquired in the second test data acquisition step into a second prediction model that uses test data from multiple points in time during hospitalization to predict the clinical outcome of the patient being predicted, thereby predicting the subsequent clinical outcome of the patient being predicted. It is a configuration that includes [this].

[0016] To solve the aforementioned problems, a method for predicting the clinical outcome of a disease according to one aspect of the present invention includes a first information processing method that predicts the subsequent clinical outcome of a patient from examination data at the initial consultation, A second information processing method for predicting the subsequent clinical outcome of a target patient from laboratory data collected during hospitalization, Includes, The first information processing method is A first training data acquisition step involves obtaining training data that includes one or more pairs of initial examination data and clinical outcomes of patients suffering from the target disease, A first learning step involves using the initial examination data as input to train a first predictive model that predicts the clinical outcome, by referring to the training data. A first test data acquisition step involves acquiring test data from the patient to be predicted at the time of their first visit, A first prediction step involves inputting the test data acquired in the first test data acquisition step into the first prediction model to predict the subsequent clinical outcome of the patient to be predicted, Includes, The second information processing method described above is: A second training data acquisition step involves obtaining training data that includes one or more pairs of laboratory data from patients suffering from the target disease during hospitalization and their clinical outcomes. A second learning step involves using test data from multiple points in time during hospitalization as input to train a second predictive model that predicts the clinical outcome, by referring to the training data. A second test data acquisition step involves acquiring test data at multiple points in time during the hospitalization of the patient to be predicted, A second prediction step involves inputting the test data acquired in the second test data acquisition step into the second prediction model to predict the subsequent clinical outcome of the patient being predicted, It is a configuration that includes [this].

[0017] To solve the aforementioned problems, a method for predicting the clinical outcome of a disease according to one aspect of the present invention includes a first information processing method that predicts the subsequent clinical outcome of a patient from examination data at the initial consultation, A second information processing method for predicting the subsequent clinical outcome of a target patient from laboratory data collected during hospitalization, Includes, The first information processing method is A first test data acquisition step involves acquiring test data from the patient to be predicted at the time of their first visit, A first prediction step predicts the subsequent clinical outcome of the patient being predicted by inputting the examination data acquired in the examination data acquisition step into a first prediction model that uses the examination data from the initial examination as input to predict the clinical outcome, Includes, The second information processing method described above is: A second test data acquisition step involves acquiring test data at multiple points in time during the hospitalization of the patient to be predicted, A second prediction step involves inputting the test data acquired in the test data acquisition step into a second prediction model that uses test data from multiple points in time during hospitalization to predict the clinical outcome of the patient being predicted, thereby predicting the subsequent clinical outcome of the patient being predicted. It is a configuration that includes [this].

[0018] To solve the aforementioned problems, a clinical outcome prediction program for a disease according to one aspect of the present invention is a program that causes an information processing device to execute a second information processing method that predicts the subsequent clinical outcome of a target patient from examination data during hospitalization, The second information processing method described above is: A training data acquisition step involves acquiring training data that includes one or more pairs of laboratory data from patients suffering from the target disease during hospitalization and their clinical outcomes. A learning step involves using test data from multiple points in time during hospitalization as input to train a predictive model that predicts the clinical outcome, by referring to the training data, A step of acquiring test data to acquire test data at multiple points in time during the hospitalization of the patient to be predicted, A prediction step in which the subsequent clinical outcome of the patient to be predicted is predicted by inputting the test data acquired in the test data acquisition step into the prediction model, It is a configuration that includes [this].

[0019] To solve the aforementioned problems, a clinical outcome prediction program for a disease according to one aspect of the present invention is a program that causes an information processing device to execute a first information processing method that predicts the subsequent clinical outcome of a target patient from examination data at the initial consultation, The first information processing method is A training data acquisition step involves acquiring training data that includes one or more pairs of initial examination data and clinical outcomes of patients suffering from the target disease, and A learning step in which a predictive model that predicts the clinical outcome is trained by taking the initial examination data as input and referring to the training data, A step of acquiring examination data to acquire examination data at the initial examination of the patient to be predicted, A prediction step in which the subsequent clinical outcome of the patient to be predicted is predicted by inputting the test data acquired in the test data acquisition step into the prediction model, It is a configuration that includes [this].

[0020] To solve the aforementioned problems, a clinical outcome prediction program for a disease according to one aspect of the present invention is a program that causes an information processing device to execute a second information processing method that predicts the subsequent clinical outcome of a target patient from examination data during hospitalization, The second information processing method described above is: A step of acquiring test data to acquire test data at multiple points in time during the hospitalization of the patient to be predicted, A prediction step in which the subsequent clinical outcome of the patient to be predicted is predicted by inputting the test data acquired in the test data acquisition step into a prediction model that predicts the clinical outcome, using test data from multiple points in time during hospitalization as input, It is a configuration that includes [this].

[0021] To solve the aforementioned problems, a clinical outcome prediction program for a disease according to one aspect of the present invention is a program that causes an information processing device to execute a first information processing method that predicts the subsequent clinical outcome of a target patient from examination data at the initial consultation, The first information processing method is A step of acquiring examination data to acquire examination data at the initial examination of the patient to be predicted, A prediction step in which the subsequent clinical outcome of the patient to be predicted is predicted by inputting the examination data acquired in the examination data acquisition step into a prediction model that predicts the clinical outcome, using the examination data from the initial examination as input, It is a configuration that includes [this].

[0022] The aforementioned program, and the computer-readable recording medium on which it is recorded, also fall within the scope of the present invention. [Effects of the Invention]

[0023] According to one aspect of the present invention, it is possible to predict changes in physical condition, including clinical outcomes, with high accuracy in various clinical situations. [Brief explanation of the drawing]

[0024] [Figure 1] This is a block diagram showing the configuration of a clinical outcome prediction system according to Embodiment 1 of the present invention. [Figure 2]This is a flowchart showing the flow of the information processing method performed by the information processing device included in the clinical outcome prediction system according to Embodiment 1 of the present invention. [Figure 3] This is a flowchart showing the flow of the information processing method performed by the information processing device included in the clinical outcome prediction system according to Embodiment 1 of the present invention. [Figure 4] This is a block diagram showing the configuration of a clinical outcome prediction system according to Embodiment 2 of the present invention. [Figure 5] This is a flowchart showing the flow of the information processing method performed by the information processing device included in the clinical outcome prediction system according to Embodiment 2 of the present invention. [Figure 6] This is a block diagram showing the configuration of an information processing device included in a clinical outcome prediction system according to Embodiment 3 of the present invention. [Figure 7] This is a flowchart showing the flow of the information processing method performed by the information processing device included in the clinical outcome prediction system according to Embodiment 3 of the present invention. [Figure 8] This is a block diagram showing the configuration of a clinical outcome prediction system according to Embodiment 4 of the present invention. [Figure 9] This is a flowchart showing the flow of the information processing method performed by the information processing device included in the clinical outcome prediction system according to Embodiment 4 of the present invention. [Figure 10] This is a block diagram showing the configuration of a clinical outcome prediction system according to Embodiment 5 of the present invention. [Figure 11] This is a block diagram showing the hardware configuration of an information processing device included in a clinical outcome prediction system according to embodiments 1 to 5 of the present invention. [Figure 12] This figure shows the screening results for severity and prognosis at the initial consultation. [Figure 13] This figure shows the results of evaluating the importance of variables in predicting each severity level of COVID-19 using SHAP values. [Figure 14] This figure shows box and jitter plots representing the distribution of important continuous variables for each severity group of COVID-19 patients. [Figure 15]This figure shows the changes in the severity status during hospitalization for COVID-19 patients, categorized by severity outcome group. [Figure 16] This figure shows the results of evaluating the mortality risk of COVID-19 patients using a 7-day cumulative hazard function (CHF) for mortality, calculated using RSF as the indicator. [Figure 17] This figure shows the results of analyzing the variables that contribute to RSF prediction. [Figure 18] This figure shows the results of confirming the agreement between RSF prediction and the actual clinical course in severe cases. [Figure 19] This figure shows the results of a comparative study, evaluating the risk of death based on LDH, lymphocytes, and CRP identified by the NMI model. [Figure 20] This figure shows the results of comparing the accuracy of the NMI model (comparative example) with the accuracy of the RSF model in this embodiment. [Figure 21] This figure shows the results of an examination of the effectiveness of the RSF model in predicting mortality and hospital discharge in non-COVID-19 cases. [Figure 22] This figure illustrates a functional method for predicting clinical outcomes from disease data, which forms the basis of the prediction program of the information processing device included in the clinical outcome prediction system according to Embodiments 1 to 5 of the present invention. [Figure 23] This is a conceptual diagram illustrating concepts for inference. [Modes for carrying out the invention]

[0025] [Outline of a functional system according to one aspect of the present invention and the problems it solves] Referring to Figure 22, the functional methods (methods, systems, or programs) for predicting clinical outcomes from disease data, which form the basis of the prediction program of the information processing device included in the clinical outcome prediction system according to Embodiments 1 to 5 of the present invention, will be explained. Hospitalized patients undergo examinations starting with an initial examination and continuing with examinations over time. Predictions of the patient's prognosis and outcome are divided into those based on initial examination data (first information processing method) and those based on examination data performed over time (second information processing method).

[0026] One aspect of the present invention enables highly accurate prediction of changes in physical condition, including clinical outcomes, in various clinical situations by executing a clinical outcome prediction program that combines programs (a first program and a second program) for executing both of these information processing methods. In the case of prediction from examination data at the initial consultation, the physical condition at the initial consultation is considered as a single state. In the case of prediction from examination data over time during hospitalization, the period is considered as a single state. In either case, differences in states are identified based on the distance between states. In one aspect of the present invention, the first information processing method using examination data at the initial consultation employs a first prediction model trained using a gradient boosting decision tree. In one aspect of the present invention, the second information processing method using time-series data during hospitalization employs a second prediction model trained using a random survival forest.

[0027] Figure 23 is a conceptual diagram showing the concept of inference by an information processing device included in the clinical outcome prediction system according to Embodiments 1 to 5 of the present invention. Figure 23 illustrates a problem to be solved by one aspect of the present invention. Knowledge in biomedical science is an integration of universal causal knowledge obtained through comparative controlled trials as a directed graph using the concepts of linear algebra. When using this knowledge to predict the state of an individual and its changes, it is necessary to comprehensively measure the molecular network between cells and construct a predictive model. However, in reality, neither such measurements nor the creation of a predictive model from measured values ​​is possible. To solve this problem, the clinical outcome of a patient who has developed a disease is predicted from a predictive system that identifies the state of an individual and the progression of that state from various test data using the method of the present invention.

[0028] On the other hand, the states (i.e., clinical outcomes) and the transitions between states identified by the information processing method according to one aspect of the present invention can be used to select drugs and therapeutic interventions that reflect the differences in states. This is because each identified state corresponds to a specific network represented by algebra, and by estimating the structure of the network from the identified states in a reverse manner than usual, it is possible to know which elements to focus on when developing drugs or therapeutic interventions. In other words, one aspect of the present invention can be used to realize personalized medicine.

[0029] [Embodiment 1] Hereinafter, one embodiment of the present invention will be described in detail. Referring to Figure 1, the clinical outcome prediction system 1000 for diseases according to Embodiment 1 (hereinafter simply referred to as "clinical outcome prediction system 1000") will be described. Figure 1 is a block diagram showing the configuration of the clinical outcome prediction system 1000 according to Embodiment 1 of the present invention. The clinical outcome prediction system 1000 according to Embodiment 1 includes an information processing device 1 that executes a first program which causes the information processing device to execute a first information processing method which predicts the subsequent clinical outcome of a target patient from examination data at the initial examination, and a second program which causes the information processing device to execute a second information processing method which predicts the subsequent clinical outcome of a target patient from examination data during hospitalization. The clinical outcome prediction system 1000 according to Embodiment 1 further includes a database 2 connected to the information processing device 1.

[0030] (Configuration of Information Processing Device 1) Referring to Figure 1, the configuration of the information processing device 1 according to Embodiment 1 will be described.

[0031] As shown in Figure 1, the information processing device 1 comprises a control unit 11, a storage unit 12, an operation reception unit 13, a display unit 14, a communication unit 15, and a memory 16.

[0032] The communication unit 15 communicates with the database 2 via the network N. More specifically, the communication unit 15 receives examination data of the patient to be predicted from the database 2. The communication unit 15 supplies the examination data of the patient to be predicted received from the database 2 to the storage unit 12. The specific configuration of the network N is not limited to this embodiment, but as an example, a wireless LAN (Local Area Network), wired LAN, WAN (Wide Area Network), public telephone network, mobile data communication network, or a combination of these networks can be used.

[0033] The memory unit 12 stores various types of data that the control unit 11 references. For example, the memory unit 12 stores the training data TD that the learning unit 112 (described later) references. It also stores the model parameters MP that define the prediction model trained by the learning unit 112. Furthermore, it stores the examination data of the patient to be predicted, received from the database 2. The memory unit 12 also stores the first program and the second program (described later).

[0034] Memory 16 contains various applications for realizing the functions of each part of the control unit 11. The control unit 11 realizes the functions of each part by writing data to and reading data from memory 16.

[0035] As shown in Figure 1, the control unit 11 includes a data acquisition unit 111, a learning unit 112, a prediction unit 113, and an output unit 114. The control unit 11 may also be described as functioning as a data acquisition unit 111, a learning unit 112, a prediction unit 113, and an output unit 114.

[0036] The control unit 11 executes the first information processing method by, for example, executing the first program. The control unit 11 also executes the second information processing method by, for example, executing the second program. In other words, the first program is a program that causes the control unit 11 to execute the first information processing method, and the second program is a program that causes the control unit 11 to execute the second information processing method.

[0037] The information processing device 1 functions as a machine learning device during the learning phase and as a predictive device during the prediction phase. The machine learning device and the predictive device are described below.

[0038] ≪Machine Learning Stage≫ This document describes an information processing device 1 as a machine learning device for training a predictive model. Note that in this specification, the expression "train a model" may also be expressed as "train a model."

[0039] The data acquisition unit 111 acquires from the storage unit 12 training data TD that includes one or more pairs of initial examination data and clinical outcomes of patients suffering from the target disease, or training data TD that includes one or more pairs of examination data and clinical outcomes of patients suffering from the target disease during hospitalization. Details about the examination data and training data TD will be described later.

[0040] The learning unit 112 trains a prediction model by referring to the training data TD acquired by the data acquisition unit 111. Then, the learning unit 112 stores the model parameters MP that define the trained prediction model in the storage unit 12. Details about the prediction model will be described later.

[0041] According to the information processing device 1 configured as described above, a predictive model for predicting the clinical outcome can be suitably trained by taking examination data from the initial consultation or examination data from multiple points in time during hospitalization as input. "Outcome" refers to the course or result of symptoms in the treatment of a disease or other condition.

[0042] The following describes the test data, training data (TD), and predictive model.

[0043] (Items included in the test data) Laboratory data refers to a dataset that includes data on at least multiple items, such as symptoms; patient background information (age, sex, etc.); biological information (medical record information, clinical laboratory data, liquid biopsy (proteome, metabolome, exosome), single-cell profile (transcriptome), diagnostic imaging data (CT, X-ray), physiological function (ECG, heart rate, EEG), psychological characteristics, genomic and epigenomic information, microbiome information, etc.); blood test data; and urine test data. The data for each item included in the laboratory data can be obtained from the electronic medical records of patients suffering from the target disease.

[0044] (Training data)

[0045] The first example of training data TD is training data that includes one or more pairs of initial examination data and clinical outcomes for patients suffering from the target disease. In the following description, this training data will be referred to as "First Training Data TD1". The method of generating First Training Data TD1 is not limited to this embodiment, but as an example, the clinical outcomes of patients suffering from the target disease are associated with the initial examination data of those patients to create pairs of initial examination data and clinical outcomes for patients suffering from the target disease. By creating these pairs for multiple patients suffering from the target disease, First Training Data TD1 containing multiple pairs of initial examination data and clinical outcomes for patients suffering from the target disease can be generated.

[0046] A second example of training data TD is training data that includes one or more pairs of laboratory data from patients suffering from the target disease during their hospitalization and their clinical outcomes. In the following description, this training data will be referred to as "second training data TD2". The method of generating second training data TD2 is not limited to this embodiment, but as an example, first, a pair of laboratory data from a patient suffering from the target disease during their hospitalization is associated with the patient's clinical outcome to create a pair of laboratory data from a patient suffering from the target disease during their hospitalization and their clinical outcome. Similarly, a pair of laboratory data from a patient suffering from the target disease during their hospitalization is associated with the patient's clinical outcome to create a pair of laboratory data from a patient suffering from the target disease during their hospitalization and their clinical outcome. By creating these pairs for multiple patients suffering from the target disease, it is possible to generate second training data TD2 that includes multiple pairs of laboratory data from patients suffering from the target disease during their hospitalization and their clinical outcomes.

[0047] Furthermore, the information processing device 1 may be configured to generate the first training data TD1 or the second training data TD2 within itself and store the generated first training data TD1 or the second training data TD2 in the storage unit 12, or it may be configured to acquire the first training data TD1 or the second training data TD2 generated outside the information processing device 1 via the communication unit 15 and store the acquired first training data TD1 or the second training data TD2 in the storage unit 12.

[0048] (Predictive model) The specific configuration of the prediction model is not limited to this embodiment, but as an example, a configuration using a gradient boosting decision tree, random forest, random survival forest, energy landscape model, time series state change model, etc. may be used. Furthermore, when data is acquired on a scale of tens of thousands to hundreds of thousands of people, a neural network such as a CNN (Convolutional Neural Network) or RNN (Recurrent Neural Network) may be used in addition to, or instead of, the above models. The specific configuration of the prediction model can be appropriately selected according to the purpose of the prediction processing by the prediction unit 113.

[0049] The first example of a predictive model is one that takes initial examination data as input and predicts the clinical outcome. In the following description, this predictive model will be referred to as the "first predictive model." The learning unit 112 can train the first predictive model by referring to the first training data TD1. A gradient boosting decision tree is a machine learning method that combines three techniques: "gradient descent," "ensemble learning," and "decision trees," and is the most powerful method among supervised learning. It is preferable that the first predictive model be trained using a gradient boosting decision tree because it does not require scaling, can prevent overfitting by adjusting the depth of the decision tree, and can identify variables important for discrimination and perceive them as existing knowledge.

[0050] A second example of a predictive model is one that takes test data from multiple points in time during hospitalization as input and predicts the clinical outcome. In the following description, this predictive model will be referred to as the "second predictive model." The learning unit 112 can train the second predictive model by referring to the second training data TD2. It is preferable that the second predictive model be trained using a random survival forest because using multiple decision trees improves prediction accuracy, adjusting the decision trees can adjust bias and variance to prevent overfitting, and variables that play an important role in prediction can be identified and compared with existing knowledge.

[0051] (Flow of machine learning method M1 using information processing device 1) Next, referring to Figure 2, we will explain the flow of the machine learning method (information processing method) M1 performed by the information processing device 1. Figure 2 is a flowchart showing the flow of the machine learning method M1 performed by the information processing device 1. The first machine learning method M1 (first information processing method) will be explained below.

[0052] (Step M11: First training data acquisition step) First, in step M11, the data acquisition unit 111 acquires the first training data TD1.

[0053] (Step M12: First Learning Step) Next, in step M12, the learning unit 112 trains the first prediction model by referring to the first training data TD1 acquired by the data acquisition unit 111. Here, the first prediction model that the learning unit 112 trains is, as described above, a first prediction model that takes the initial examination data of the patient to be predicted as input and predicts the clinical outcome.

[0054] As an example, the learning unit 112 inputs the test data included in the first training data TD1 acquired by the data acquisition unit 111 into the first prediction model. The learning unit 112 then updates the parameters of the first prediction model so that the difference between the clinical outcome associated with the test data and the clinical outcome output by the first prediction model is minimized. The learning unit 112 trains the first prediction model by repeatedly performing this update process. The parameters of the first prediction model after training are stored in the storage unit 12 as the model parameters MP1 described above.

[0055] According to the processing performed by the information processing device 1 described above, a first predictive model for predicting the clinical outcome can be suitably trained using the initial examination data of the patient to be predicted as input.

[0056] Furthermore, in step M11 of the aforementioned machine learning method M1, the second machine learning method M1' (second information processing method) can be performed by replacing the training data acquired by the data acquisition unit 111 with the second training data TD2. In the second machine learning method M1', in step M11 (second training data acquisition step), the data acquisition unit 111 acquires the second training data TD2. In step M12 (second learning step), the learning unit 112 learns the second predictive model by referring to the second training data TD2. Through this processing by the information processing device 1, a second predictive model that predicts the clinical outcome can be suitably learned by taking examination data from multiple points in time during the hospitalization of the patient to be predicted as input.

[0057] <Prediction Stage> An information processing device 1, which predicts the subsequent clinical outcome of a patient based on their inpatient examination data or initial examination data, will be explained with reference to Figure 1.

[0058] The data acquisition unit 111 acquires the examination data D of the patient to be predicted from the storage unit 12.

[0059] The prediction unit 113 performs a prediction process to predict the subsequent clinical outcome of a patient by inputting the examination data D of the patient to be predicted, acquired by the data acquisition unit 111, into the prediction model trained by the learning unit 112. Details about the examination data D and the prediction results will be described later.

[0060] The output unit 114 outputs the prediction result from the prediction unit 113 to the display unit 14.

[0061] The display unit 14 presents the clinical outcome based on the prediction results output by the output unit 114. The display unit 14 may be equipped with a display panel and configured to present the clinical outcome based on the prediction results from the prediction unit 113 as an image. Alternatively, the display unit 14 may be equipped with a speaker and configured to present the clinical outcome based on the prediction results from the prediction unit 113 as sound. With the information processing device 1 equipped with the display unit 14 as described above, the clinical outcome based on the prediction results from the prediction unit 113 can be suitably notified to the operator.

[0062] In this way, the information processing device 1 uses the patient's examination data and both the first and second prediction models for the target disease to predict the subsequent clinical outcome of the patient. Therefore, it can accurately predict changes in physical condition, including clinical outcomes, in a variety of clinical situations.

[0063] The following describes the inspection data D used by the prediction unit 113 for prediction processing and the prediction results generated by the prediction unit 113 through prediction processing.

[0064] (Inspection data D) The items included in the test data D are as described above. The first example of test data D is the test data from the initial consultation of the patient to be predicted. In the following explanation, this test data will be referred to as "first test data D1". The data for each item included in test data D1 can be obtained from the electronic medical record of the patient to be predicted at the initial consultation. Depending on the purpose of the prediction processing by the prediction unit 113, a combination of data for the necessary items should be selected and input as test data D1 into the first prediction model described above. For example, as test data D1, test data including at least 84 items of data such as background information and blood / urine test data shown in Table 1 may be used. [Table 1] JPEG0007839538000002.jpg213163JPEG0007839538000003.jpg217162JPEG0007839538000004.jpg220163JPEG00078395380 00005.jpg218163JPEG0007839538000006.jpg218163JPEG0007839538000007.jpg219163JPEG0007839538000008.jpg161162

[0065] A second example of test data D is test data from multiple points in time during the hospitalization of the patient being predicted. In the following explanation, this test data will be referred to as "second test data D2". The data for each item included in test data D2 can be obtained from the electronic medical record of the patient being predicted at any time during hospitalization. Depending on the purpose of the prediction processing by the prediction unit 113, the necessary combination of data items should be selected and input into the second prediction model described above as test data D2. For example, as test data D2, test data including 93 items such as blood tests and vital information shown in Table 2 may be used. [Table 2] JPEG0007839538000010.jpg210162JPEG0007839538000011.jpg204162JPEG000 7839538000012.jpg207162JPEG0007839538000013.jpg208164JPEG0007839538 000014.jpg211162JPEG0007839538000015.jpg206163JPEG0007839538000016. jpg207163JPEG0007839538000017.jpg210162JPEG0007839538000018.jpg90162

[0066] It should be noted that it is rare for all the items exemplified to be measured in all patients, and the types of items tested vary from patient to patient as needed. In such cases, it is sufficient to select variables for handling missing data and perform substitution processing for missing values ​​(missing value substitution processing).

[0067] In the missing value substitution process, a criterion may be set for a certain measurement item to be missing more than a certain percentage (for example, 20% or more). If more than the predetermined percentage is missing, that item should be excluded from the test data. Then, for the remaining items, the missing parts can be filled in using a known substitution method. Here, as a known substitution method, a single substitution method using a random forest, such as missForest, may be used, or it may be used in combination with a multiple substitution method that repeats substitution multiple times. The complete test data created in this way, with no missing items, can then be used for prediction processing by the prediction unit 113.

[0068] (Prediction results) The prediction results from the prediction unit 113 include at least information indicating which of the multiple clinical outcomes of the target disease the patient being predicted to reach will be. Here, the "multiple clinical outcomes of the target disease" only needs to include at least two clinical outcomes. Although not limiting to this embodiment, as an example, the multiple clinical outcomes of the target disease may include two types of clinical outcomes corresponding to cure and death. As another example, as shown in the embodiment described later, the multiple clinical outcomes of the target disease may include six types of clinical outcomes classified based on the oxygen support status score.

[0069] The types of "target diseases" mentioned above are not particularly limited and can be selected as appropriate depending on the purpose. The clinical outcome prediction system 1000 according to Embodiment 1 is applicable to infectious diseases, lifestyle-related diseases (non-communicable diseases) in general, and mental illnesses.

[0070] In this specification, "infectious disease" refers to acute respiratory illness caused by the SARS-CoV-2 virus (COVID-19), bacterial pneumonia, bacterial prostatitis, urinary tract infections, Ebola hemorrhagic fever, Crimean-Congo hemorrhagic fever, smallpox, South American hemorrhagic fever, Marburg disease, Lassa fever, plague, polio, tuberculosis, diphtheria, severe acute respiratory syndrome (SARS), Middle East respiratory syndrome (MERS), avian influenza (including but not limited to H5N1 and H7N9), cholera, enterohemorrhagic Escherichia coli infection, bacterial dysentery, Typhoid fever, paratyphoid fever, hepatitis E, hepatitis A, echinococcosis, yellow fever, psittacosis, relapsing fever, Q fever, coccidioidomycosis, Zika virus infection, tick-borne encephalitis, anthrax, scrub typhus, botulism, malaria, tularemia, Lyme disease, Rift Valley fever, meridianus, legionellosis, West Nile fever, rabies, severe fever with thrombocytopenia syndrome (SFTS), chikungunya fever, dengue fever, Japanese spotted fever, Japanese encephalitis, hantavirus pulmonary syndrome, brucellosis, leptospirosis, Rocky Mountain spotted fever, amoebic dysentery, viral hepatitis, carbapenem-resistant Enterobacteriaceae Bacterial infections, acute encephalitis, cryptosporidiosis, Creutzfeldt-Jakob disease, fulminant hemolytic streptococcal infection, acquired immunodeficiency syndrome (HIV / AIDS), giardiasis, invasive Haemophilus influenzae infection, invasive meningococcal infection, invasive pneumococcal infection, congenital rubella syndrome (CRS), syphilis, disseminated cryptococcosis, tetanus, vancomycin-resistant Staphylococcus aureus infection, vancomycin-resistant enterococcal infection, pertussis, rubella, measles, drug-resistant Acinetobacter infection, respiratory syncytial virus (RSV) infection, pharyngoconjunctival fever (pool fever), group A hemolysis It may be any of the following infections: sexual streptococcal pharyngitis, infectious gastroenteritis (including but not limited to norovirus and rotavirus), chickenpox, hand, foot, and mouth disease, erythema infectiosum, roseola, herpangina, mumps, influenza, acute hemorrhagic conjunctivitis, epidemic keratoconjunctivitis, chlamydial pneumonia, bacterial meningitis, mycoplasma pneumonia, aseptic meningitis, genital chlamydia infection, genital herpesvirus infection, genital warts, gonorrhea, penicillin-resistant pneumococcal infection, methicillin-resistant Staphylococcus aureus infection, drug-resistant Pseudomonas aeruginosa infection, and any other infection.In addition to these known infectious diseases, emerging infectious diseases caused by viruses or bacteria that have not yet been identified or do not currently exist are also included in the scope of infectious diseases as defined herein. The “infectious disease” is preferably an acute infectious disease.

[0071] In this specification, "lifestyle-related diseases" refers to a group of diseases in which lifestyle habits such as eating habits, exercise habits, rest, smoking, and drinking are involved in the onset and progression of the disease. Examples include, but are not limited to, cancer (malignant neoplasms, carcinoma in situ, etc.), heart disease, cerebrovascular disease, diabetes, hypertensive disease, cirrhosis of the liver, and chronic renal failure.

[0072] In this specification, "mental illness" refers to a disorder resulting from functional impairment or organic problems of the brain, and can also be referred to as a mental disorder. Examples include, but are not limited to, mood disorders such as depression, schizophrenia, and dementia.

[0073] The information included in the prediction results from the prediction unit 113 is not limited to this embodiment, but the prediction results may be configured to include the probability that the patient being predicted will have the clinical outcome indicated in the prediction results. Here, this probability may also be expressed as accuracy.

[0074] (Flowchart of Clinical Outcome Prediction Method M2 using Information Processing Device 1) Next, referring to Figure 3, the flow of the clinical outcome prediction method (information processing method) M2 by the information processing device 1 will be explained. Figure 3 is a flowchart showing the flow of the clinical outcome prediction method M2 by the information processing device 1. The information processing device 1 executes the clinical outcome prediction method M2. The second clinical outcome prediction method M2 (second information processing method) will be explained below.

[0075] The second clinical outcome prediction method M2 includes a second test data acquisition step M21 for acquiring test data D2 at multiple points in time during the hospitalization of the patient to be predicted; a second prediction step M22 for predicting the subsequent clinical outcome of the patient to be predicted by inputting the test data D2 acquired in the second test data acquisition step to the second prediction model; and a second output step M23 for outputting the prediction result from the second prediction step.

[0076] (Step M21: Second test data acquisition step) First, in step M21, the data acquisition unit 111 acquires examination data D2 from multiple points in time during the hospitalization of the patient to be predicted, which is stored in the storage unit 12.

[0077] The test data D2 can be obtained at multiple points in time during the observation period (during hospitalization), and there are no particular restrictions on when the test data is obtained at these multiple points in time. However, there is a discrepancy between the timing of obtaining the test data and the timing of onset or worsening of the disease, and each patient follows a different course. Furthermore, there are cases where patients who showed no signs of worsening at the time of admission suddenly become severely ill and die during hospitalization. Since the risk of severe illness and death changes dynamically even within an individual, the shorter the time interval between obtaining new test data, the more accurately the outcome can be predicted. However, considering the effort involved in obtaining the test data and the burden on the patient, it is usually sufficient to obtain the test data at 3-day intervals, preferably at 1-day intervals.

[0078] (Step M22: Second prediction step) Next, in step M22, the prediction unit 113 inputs the examination data D2 acquired by the data acquisition unit 111 at multiple points in time during the patient's hospitalization into the second prediction model trained by the learning unit 112, thereby performing a prediction process to predict the subsequent clinical outcome of the patient.

[0079] (Step M23: Second output step) Next, in step M23, the output unit 114 outputs the prediction results from the prediction unit 113 to the display unit 14. The prediction results obtained through the prediction process show the mortality risk (clinical outcome) at multiple time points. This makes it possible to perform a dynamic assessment of mortality risk.

[0080] Thus, by processing by the information processing device 1, the subsequent clinical outcome of the patient is predicted by inputting examination data D2 at multiple points in time during the patient's hospitalization into the second prediction model for the target disease. Therefore, the clinical outcome of the target disease for the patient can be predicted with high accuracy.

[0081] Furthermore, in step M21 of the aforementioned clinical outcome prediction method M2, the first clinical outcome prediction method M2' (first information processing method) can be performed by replacing the test data acquired by the data acquisition unit 111 with the first test data D1. In the first clinical outcome prediction method M2', in step M21 (first test data acquisition step), the data acquisition unit 111 acquires the test data D1 of the patient to be predicted at the time of the initial examination, which is stored in the memory unit 12. In step M22 (first prediction step), the prediction unit 113 inputs the test data D1 of the patient to be predicted at the time of the initial examination, acquired by the data acquisition unit 111, into the second prediction model trained by the learning unit 112, thereby executing a prediction process to predict the subsequent clinical outcome of the patient to be predicted. The prediction result obtained by the prediction process shows the severity prognosis at the time of the initial examination. This makes it possible to perform a severity screening at the time of the initial examination based on the prediction result, and since it is possible to predict the outcome of whether or not to be hospitalized and whether or not the condition will worsen after hospitalization at an early stage such as the initial examination, it is possible to screen for estimation of the priority allocation of resources in clinical practice.

[0082] Processing by the information processing device 1 allows for highly accurate prediction of changes in physical condition, including clinical outcomes, in various clinical situations. This is expected to lead to appropriate interventions for each patient and the prior preparation of optimal medical resources. This will contribute to achieving Sustainable Development Goal (SDG) 3: "Ensure healthy lives and promote well-being for all."

[0083] (Additional information for Embodiment 1) In the above description, an example was given of a configuration in which the information processing device 1 performs a clinical outcome prediction method by combining both the first information processing method and the second information processing method, but this is not limited to this embodiment. The information processing device 1 may be configured to perform at least one of the first information processing method or the second information processing method.

[0084] [Embodiment 2] Other embodiments of the present invention are described below. For the sake of clarity, components having the same function as those described in the above embodiments will be denoted by the same reference numerals, and their descriptions will not be repeated.

[0085] Figure 4 is a block diagram showing the configuration of the clinical outcome prediction system 1000a according to this embodiment. The clinical outcome prediction system 1000a according to Embodiment 2 includes an information processing device 1a. The clinical outcome prediction system 1000a according to Embodiment 2 further includes a database 2 connected to the information processing device 1a.

[0086] As shown in Figure 4, the information processing device 1a includes a specific unit 115 and a proposal unit 116 in addition to the configurations of the information processing device 1 according to Embodiment 1. Of the configurations of the information processing device 1a, the configurations other than the specific unit 115 and the proposal unit 116 are the same as those of the information processing device 1 according to Embodiment 1, so their explanation is omitted here.

[0087] The identification unit 115 is configured to identify, in the clinical outcome prediction process in the prediction step, the item among the multiple items included in the examination data that contributes more to the clinical outcome prediction process.

[0088] The proposal unit 116 is configured to propose candidate target molecules for therapeutic drugs for the target disease based on the identification results identified by the identification unit 115.

[0089] (Flow of the identification and proposal method M3 by the information processing device 1a) Next, with reference to Figure 5, the flow of the identification and proposal method (information processing method) M3 by the information processing device 1a will be explained. Figure 5 is a flowchart showing the flow of the identification and proposal method M3 by the information processing device 1a. The first information processing method and the second information processing method include the identification and proposal method M3. The information processing device 1a executes the identification and proposal method M3. The second identification and proposal method M3 (second information processing method) will be explained below.

[0090] The second identification and proposal method M3 includes an identification step M31 in which, in the clinical outcome prediction process in the prediction step M22, an item among a plurality of items included in the examination data that contributes more to the prediction process of the clinical outcome, and a proposal step M32 in which, based on the identification results of the identification step M31, a candidate target molecule for a therapeutic drug for the target disease is proposed.

[0091] (Step M31: Second Identification Step) First, in step M31, the identification unit 115 performs an identification process to identify, in the clinical outcome prediction process performed by the prediction unit 113, the item among the multiple items included in the examination data that contributes more to the prediction process of the clinical outcome.

[0092] The identification process performed by the identification unit 115 is a process of selecting items corresponding to features with relatively large weights from among the features calculated by the second prediction model described above. The specific method of the identification process is not limited to this embodiment, but as an example, by evaluating the variable importance in the second prediction model using SHAP (Shapley Additive exPlanations) values, features with large weights can be identified in the second prediction model described above. In step M31, the principle that affects clinical outcomes can be identified from the second prediction model, thereby ensuring the validity (reliability) of the predictions of the information processing device 1a according to this embodiment.

[0093] (Step M32: Second Proposal Step) Next, in step M32, the proposal unit 116 generates proposal information suggesting candidate target molecules for therapeutic drugs for the target disease, based on the identification results identified by the identification unit 115. The specific content of the proposal information is not limited to this embodiment, but as an example, it can be a list of target molecules with high variable importance in the second prediction model from the identification results identified by the identification unit 115, arranged in descending order of variable importance. This allows the user to use the proposal information to search for target molecules for therapeutic drugs for the target disease.

[0094] Thus, processing by the information processing device 1a allows for the identification of principles influencing clinical outcomes from the second prediction model, thereby ensuring the validity (reliability) of the predictions made by the information processing device 1a according to this embodiment. Furthermore, by using the suggested information from the suggestion unit 116 as a basis for the user to search for target molecules for therapeutic drugs of the target disease, it becomes possible to discover target molecules for therapeutic drugs.

[0095] Similarly, the first information processing method includes the first identification and proposal method M3'. The first identification and proposal method M3' can be performed in the same way as the second identification and proposal method M3 described above, but it differs from the second identification and proposal method M3' in that in step M31 (first identification step), it identifies the principle that influences clinical outcomes from the first prediction model. In step M32 (first proposal step), the proposal unit 116 generates proposal information that proposes candidate target molecules for therapeutic drugs for the target disease based on the identification results identified by the identification unit 115. In this way, processing by the information processing device 1a makes it possible to identify the principle that influences clinical outcomes from the first prediction model.

[0096] (Additional information for Embodiment 2) In the above description, an example was given of a configuration in which the control unit 11a includes both the specific unit 115 and the proposal unit 116. However, this is not limited to this embodiment, and the control unit 11a may be configured without the proposal unit 116.

[0097] In this configuration, the identification result identified by the identification unit 115 is directly supplied to the output unit 114. The output unit 114 then outputs the identification result from the identification unit 115 to the display unit 14. Since the identification unit 115 can identify the principles that affect clinical outcomes from the first prediction model or the second prediction model, the validity (reliability) of the predictions of the information processing device 1a according to this embodiment is ensured.

[0098] [Embodiment 3] Other embodiments of the present invention are described below. For the sake of clarity, components having the same function as those described in the above embodiments will be denoted by the same reference numerals, and their descriptions will not be repeated.

[0099] Figure 6 is a block diagram showing the configuration of the clinical outcome prediction system 1000b according to this embodiment. The clinical outcome prediction system 1000b according to Embodiment 3 includes an information processing device 1b. The clinical outcome prediction system 1000b according to Embodiment 3 further includes a database 2 connected to the information processing device 1b.

[0100] As shown in Figure 6, the information processing device 1b includes a specific unit 115 and an extraction unit 117 in addition to the configurations of the information processing device 1 according to Embodiment 1. Of the configurations of the information processing device 1a, the configurations other than the specific unit 115 and the extraction unit 117 are the same as those of the information processing device 1 according to Embodiment 1, so their explanation is omitted here. Also, the configuration of the specific unit 115 is the same as that of the specific unit 115 of the information processing device 1b according to Embodiment 2, so its explanation is omitted here.

[0101] The extraction unit 117 is configured to extract inspection data to be used in the prediction step from the inspection data acquired by the data acquisition unit 111, based on the identification result by the identification unit 115.

[0102] (Flowchart of identification and extraction method M4 by information processing device 1b) Next, with reference to Figure 7, the flow of the identification and extraction method (information processing method) M4 by the information processing device 1b will be explained. Figure 7 is a flowchart showing the flow of the identification and extraction method M4 by the information processing device 1b. The first information processing method and the second information processing method include the identification and extraction method M4. The information processing device 1b executes the identification and extraction method M4. The second identification and extraction method M4 (second information processing method) will be explained below.

[0103] The second identification and extraction method M4 includes an identification step M41 in which, in the clinical outcome prediction process in the prediction step M22, an item is identified from among a plurality of items included in the test data that contributes more to the clinical outcome prediction process, and a test data extraction step M42 in which, based on the identification result of the identification step M41, an item of test data to be used in the prediction step M22 is extracted from the test data acquired in the test data acquisition step M21.

[0104] (Step M41: Second Identification Step) Step M41 is the same as step M31 mentioned above, so the explanation will be omitted.

[0105] (Step M42: Second Inspection Data Extraction Step) Next, in step M42, the extraction unit 117 performs an extraction process to extract items of the inspection data to be used by the prediction unit 113 in the prediction step M22 from the inspection data acquired by the data acquisition unit 111, based on the identification result identified by the identification unit 115.

[0106] The extraction process of test data by the extraction unit 117 is a process in which, in the second prediction model described above, items corresponding to features with relatively large weights among the features calculated by the model are extracted from the test data used by the prediction unit 113. The specific method of the identification process is not limited to this embodiment, but as an example, any number of important items with high variable importance in the second prediction model from the identification results identified by the identification unit 115 can be extracted from the test data in order of decreasing variable importance, and the test data containing these important items can be used by the prediction unit 113 in prediction step M22. As a result, in prediction step M22, clinical outcomes are predicted based on the test data containing items with higher variable importance, making it possible to make predictions with higher accuracy.

[0107] Similarly, the first information processing method includes the first identification and extraction method M4'. The first identification and extraction method M4' performs the same processing as the second identification and extraction method M4 described above, but differs from the second identification and extraction method M44 in that, in step M42 (first inspection data extraction step), items corresponding to features with relatively large weights among the features calculated by the first prediction model are extracted from the inspection data used by the prediction unit 113. In this way, processing by the information processing device 1b makes it possible to extract important items with high variable importance in the first prediction model.

[0108] In step M42, the number of items extracted by the extraction unit 117 from the test data is not particularly limited, but from the viewpoint of capturing the diverse characteristics of the disease, it is preferable that the number of items extracted by the extraction unit 117 from the test data be 10 or more. On the other hand, from the viewpoint of preventing overfitting due to increasing the number of variables too much and reducing the burden of measuring a large number of items in the clinical setting, it is preferable that the number of items extracted by the extraction unit 117 from the test data be 20 or less. Therefore, in step M42, it is preferable that the number of items extracted by the extraction unit 117 from the test data be the top 10 to 20 items in order of variable importance in the first prediction model or the second prediction model.

[0109] The extraction unit 117 may extract from the test data (training data) used for updating (retraining) the first or second prediction model the test data or items corresponding to features with relatively large weights among the features calculated by the first or second prediction model, and use these as the test data or items used by the learning unit 112 in learning step M12. This allows the first or second prediction model to be retrained in learning step M12 based on variables with higher variable importance, thereby enabling more accurate predictions.

[0110] Thus, the processing by the information processing device 1b allows for the identification of principles influencing clinical outcomes from either the first or second prediction model, thereby ensuring the validity (reliability) of the predictions made by the information processing device 1b according to this embodiment. Furthermore, by having the prediction unit 113 perform prediction processing using the examination data extracted by the extraction unit 117, more accurate predictions become possible. [Embodiment 4] Other embodiments of the present invention are described below. For the sake of clarity, components having the same function as those described in the above embodiments will be denoted by the same reference numerals, and their descriptions will not be repeated.

[0111] Figure 8 is a block diagram showing the configuration of the clinical outcome prediction system 1000c according to this embodiment. The clinical outcome prediction system 1000c according to Embodiment 4 includes an information processing device 1c. The clinical outcome prediction system 1000c according to Embodiment 4 further includes a database 2 connected to the information processing device 1c.

[0112] As shown in Figure 8, the information processing device 1c includes a proposal unit 116 in addition to the configurations provided by the information processing device 1 according to Embodiment 1. Of the configurations provided by the information processing device 1c, the configurations other than the proposal unit 116 are the same as those provided by the information processing device 1 according to Embodiment 1, so their explanation is omitted here.

[0113] In the information processing device 1c, the proposal unit 116 is configured to propose a drug or therapeutic intervention method for a patient to be predicted, based on the first prediction result R1 and the second prediction result R2 predicted by the prediction unit 113 by executing the first information processing method and the second information processing method.

[0114] The first prediction result R1 is a prediction result obtained by the prediction unit 113 performing a first information processing method that predicts the subsequent clinical outcome of the patient based on the examination data at the initial consultation. In other words, the first prediction result R1 shows the clinical outcome of the patient based on the examination data at the initial consultation.

[0115] Furthermore, the second prediction result R2 is a prediction result obtained by the prediction unit 113 performing a second information processing method that predicts the subsequent clinical outcome of the patient being predicted from the examination data during hospitalization. In other words, the second prediction result R2 shows the clinical outcome of the patient being predicted based on the examination data during hospitalization. The first prediction result R1 and the second prediction result R2 are stored in the storage unit 12a.

[0116] (Flow of the third information processing method M5 by the information processing device 1c) Next, with reference to Figure 9, the flow of the third information processing method M5 by the information processing device 1c will be explained. Figure 9 is a flowchart showing the flow of the third information processing method M5 by the information processing device 1c. A method for predicting the clinical outcome of a disease according to one aspect of the present invention includes the first information processing method and the second information processing method described above, in addition to the third information processing method M5. The information processing device 1c executes a third program that executes the third information processing method M5. That is, a clinical outcome prediction program according to one aspect of the present invention includes a first program, a second program, and a third program. The third information processing method M5 includes a prediction result acquisition step M51 in which the data acquisition unit 111 acquires a first prediction result R1 and a second prediction result R2 (the clinical outcome of the patient predicted by the first information processing method and the second information processing method) stored in the storage unit 12a, a proposal step M52 in which a drug or therapeutic intervention method is proposed based on the first prediction result and the second prediction result, and an output step M53 in which the proposed result is output.

[0117] (Step M51) First, in step M51, the data acquisition unit 111 acquires the first prediction result R1 and the second prediction result R2 stored in the storage unit 12a.

[0118] (Step M52: Proposal Step) Next, in step M52, the proposal unit 116 performs a proposal process to suggest a drug or therapeutic intervention method for the patient being predicted, referring to the first prediction result R1 and the second prediction result R2 acquired by the data acquisition unit 111.

[0119] The proposal processing by the proposal unit 116 involves referring to the first prediction result R1 and the second prediction result R2 to identify differences in the patient's condition (clinical outcome) and selecting and proposing appropriate drugs or therapeutic interventions based on the differences in condition and the progression of the condition. This allows for the optimization of the drug effect or therapeutic intervention method for the patient by referring to the first prediction result R1 and the second prediction result R2, which are predicted with high accuracy by the prediction unit 113.

[0120] The types of drugs or therapeutic interventions proposed by the proposal unit 116 are not particularly limited. The proposal unit 116 can propose the drug or therapeutic intervention that is most appropriate for the patient's condition (clinical outcome) from among all conceivable drug or therapeutic intervention options.

[0121] (Step M53) Next, in step M53, the output unit 114 outputs the proposal result from the proposal unit 116 to the display unit 14.

[0122] Thus, by processing by the information processing device 1c, the first prediction result R1 and the second prediction result R2, which are predicted with high accuracy by the prediction unit 113, can be referenced to propose the drug or therapeutic intervention method that is most suitable for the patient's condition (clinical outcome). This makes it possible to optimize the effect of the drug or therapeutic intervention method for the patient. In other words, the clinical outcome prediction system 1000c according to this embodiment can be applied to the development of personalized medicine.

[0123] [Embodiment 5] Other embodiments of the present invention are described below. For the sake of clarity, components having the same function as those described in the above embodiments will be denoted by the same reference numerals, and their descriptions will not be repeated.

[0124] Figure 10 is a block diagram showing the configuration of the clinical outcome prediction system 1000d according to this embodiment. The clinical outcome prediction system 1000d according to Embodiment 5 includes an information processing device 1d. The clinical outcome prediction system 1000d according to Embodiment 5 further includes a database 2 connected to the information processing device 1d.

[0125] As shown in Figure 10, the information processing device 1d is configured in a way that the learning unit 112 is not included in the information processing device 1 according to Embodiment 1. Since the configurations of the information processing device 1d are the same as those of the information processing device 1 according to Embodiment 1, their explanation is omitted here.

[0126] The information processing device 1d, configured as described above, functions as a prediction device. The information processing device 1d uses the initial examination data D1 of the patient to be predicted and the aforementioned first prediction model for the target disease to predict the subsequent clinical outcome of the patient. Furthermore, the information processing device 1d uses the examination data D2 at multiple points in time during the patient's hospitalization and the aforementioned second prediction model for the target disease to predict the subsequent clinical outcome of the patient. By combining both of these programs, the information processing device 1d can predict changes in physical condition, including clinical outcomes, with high accuracy in various clinical situations.

[0127] (Additional information for Embodiment 5) The information processing device 1d may perform a third information processing method, which includes the step of proposing a drug or therapeutic intervention method based on the clinical outcome of the patient predicted by the first and second information processing methods. For this purpose, the information processing device 1d may further include a proposal unit 116. This allows the information processing device 1d to refer to the first prediction result R1 and the second prediction result R2 to identify differences in the patient's condition (clinical outcome) and to select and propose an appropriate drug or therapeutic intervention method based on the differences in condition and the progression of the condition.

[0128] [Examples of implementation using software] The functions of the information processing devices 1, 1a, 1b, 1c, and 1d (hereinafter referred to as "devices") can be realized by programs that cause the devices to function as computers, and by programs that cause each control block of the devices (particularly the parts included in the control units 11, 11a, 11b, 11c, and 11d) to function as computers.

[0129] Devices 1, 1a, 1b, 1c, and 1d execute a first information processing method by executing a first program and execute a second information processing method by executing a second program.

[0130] In this case, the device includes a computer having at least one control device (e.g., a processor) and at least one storage device (e.g., memory) as hardware for executing the program. By executing the program using this control device and storage device, the functions described in each of the embodiments are realized.

[0131] The above program may be recorded on one or more computer-readable recording media, not temporary ones. These recording media may or may not be provided by the above device. In the latter case, the program may be supplied to the above device via any wired or wireless transmission medium.

[0132] Furthermore, some or all of the functions of each of the above control blocks can also be realized by logic circuits. For example, an integrated circuit in which logic circuits functioning as each of the above control blocks are formed is also included in the scope of the present invention. In addition, it is also possible to realize the functions of each of the above control blocks by, for example, a quantum computer.

[0133] The present invention is not limited to the embodiments described above, and various modifications are possible within the scope of the claims. Embodiments obtained by appropriately combining the technical means disclosed in different embodiments are also included in the technical scope of the present invention.

[0134] 〔summary〕 A clinical outcome prediction system for a disease according to Embodiment 1 of the present invention comprises an information processing device that executes a first program which causes an information processing device to execute a first information processing method which predicts the subsequent clinical outcome of a target patient from examination data at the time of initial examination, and a second program which causes an information processing device to execute a second information processing method which predicts the subsequent clinical outcome of a target patient from examination data during hospitalization, wherein the first information processing method comprises a first training data acquisition step which acquires training data which includes one or more pairs of examination data at the time of initial examination of a patient suffering from a target disease and a clinical outcome, a first learning step which uses the examination data at the time of initial examination as input and trains a first prediction model which predicts the clinical outcome by referring to the training data, a first examination data acquisition step which acquires examination data at the time of initial examination of the target patient, and a first prediction step which predicts the subsequent clinical outcome of the target patient by inputting the examination data acquired in the first examination data acquisition step to the first prediction model, The second information processing method includes a second training data acquisition step of acquiring training data that includes one or more pairs of test data and clinical outcomes of a patient suffering from a target disease during hospitalization; a second learning step of training a second prediction model that predicts the clinical outcome by taking test data at multiple points in time during hospitalization as input and referring to the training data; a second test data acquisition step of acquiring test data at multiple points in time during hospitalization of the patient to be predicted; and a second prediction step of predicting the subsequent clinical outcome of the patient to be predicted by inputting the test data acquired in the second test data acquisition step to the second prediction model.

[0135] A clinical outcome prediction system for a disease according to aspect 2 of the present invention comprises an information processing device that executes a first program which causes an information processing device to execute a first information processing method which predicts the subsequent clinical outcome of a target patient from examination data at the time of the initial examination, and a second program which causes an information processing device to execute a second information processing method which predicts the subsequent clinical outcome of a target patient from examination data during hospitalization, wherein the first information processing method includes a first examination data acquisition step of acquiring examination data at the time of the target patient's initial examination, and a first prediction step of predicting the subsequent clinical outcome of the target patient by inputting the examination data acquired in the examination data acquisition step to a first prediction model which predicts the clinical outcome using the examination data at the time of the initial examination as input, and the second information processing method includes a second examination data acquisition step of acquiring examination data at multiple points in time during the hospitalization of the target patient, and a second prediction step of predicting the subsequent clinical outcome of the target patient by inputting the examination data acquired in the examination data acquisition step to a second prediction model which predicts the clinical outcome using the examination data at multiple points in time during hospitalization as input,

[0136] In the disease clinical outcome prediction system according to aspect 3 of the present invention, in aspect 1 or 2, it is preferable that the second prediction model is trained using a random survival forest.

[0137] In the disease clinical outcome prediction system according to aspect 4 of the present invention, in aspect 1 or 2, it is preferable that the first prediction model is trained using a gradient boosting decision tree.

[0138] A clinical outcome prediction system for a disease according to aspect 5 of the present invention, in any one of aspects 1 to 4, the first information processing method further includes a first identification step of identifying an item among a plurality of items included in the examination data that contributes more to the clinical outcome prediction process in the first prediction step, and the second information processing method may further include a second identification step of identifying an item among a plurality of items included in the examination data that contributes more to the clinical outcome prediction process in the second prediction step.

[0139] A clinical outcome prediction system for a disease according to aspect 6 of the present invention, in aspect 5, further comprises a first information processing method which proposes a candidate target molecule for a therapeutic drug for the target disease based on the identification result of the first identification step, and a second information processing method which further comprises a second proposal step which proposes a candidate target molecule for a therapeutic drug for the target disease based on the identification result of the second identification step.

[0140] A clinical outcome prediction system for a disease according to embodiment 7 of the present invention, in embodiment 5, further includes a first information processing method which extracts test data to be used in the first prediction step from the test data acquired in the first test data acquisition step based on the identification result of the first identification step, and a second information processing method which may further include a second test data extraction step which extracts test data to be used in the second prediction step from the test data acquired in the second test data acquisition step based on the identification result of the second identification step.

[0141] In the clinical outcome prediction system for diseases according to aspect 8 of the present invention, in aspect 7, it is preferable that in the first and second test data extraction steps, the top 10 to 20 items are extracted from the test data in order of their variable importance in the first or second prediction model.

[0142] The clinical outcome prediction system for a disease according to aspect 9 of the present invention may be configured such that, in any one of aspects 1 to 8, the information processing device further executes a third program which causes the information processing device to execute a third information processing method which includes the step of proposing a drug or therapeutic intervention method based on the clinical outcome of the patient predicted by the first information processing method and the second information processing method.

[0143] A clinical outcome prediction program for a disease according to aspect 10 of the present invention includes: a first program that causes an information processing device to execute a first information processing method for predicting the subsequent clinical outcome of a target patient from initial examination data; and a second program that causes an information processing device to execute a second information processing method for predicting the subsequent clinical outcome of a target patient from examination data during hospitalization, wherein the first information processing method includes: a first training data acquisition step of acquiring training data that includes one or more pairs of initial examination data of a patient suffering from a target disease and a clinical outcome; a first learning step of training a first prediction model that takes the initial examination data as input and predicts the clinical outcome by referring to the training data; a first examination data acquisition step of acquiring initial examination data of the target patient; and the first examination data acquisition step of the first prediction model. The second information processing method is a program that includes: a first prediction step of predicting the subsequent clinical outcome of the target patient by inputting the test data acquired in the program; a second training data acquisition step of acquiring training data that includes one or more pairs of test data and clinical outcomes of a patient suffering from a target disease during hospitalization; a second learning step of training a second prediction model that predicts the clinical outcome by taking test data at multiple points in time during hospitalization as input and referring to the training data; a second test data acquisition step of acquiring test data at multiple points in time during hospitalization of the target patient; and a second prediction step of predicting the subsequent clinical outcome of the target patient by inputting the test data acquired in the second test data acquisition step to the second prediction model.

[0144] A clinical outcome prediction program for a disease according to aspect 11 of the present invention includes: a first program that causes an information processing device to execute a first information processing method for predicting the subsequent clinical outcome of a patient to be predicted from examination data at the time of the initial examination; and a second program that causes an information processing device to execute a second information processing method for predicting the subsequent clinical outcome of a patient to be predicted from examination data during hospitalization, wherein the first information processing method includes: a first examination data acquisition step for acquiring examination data at the time of the patient to be predicted at the time of the initial examination; and a first prediction step for predicting the subsequent clinical outcome of the patient to be predicted by inputting the examination data acquired in the examination data acquisition step to a first prediction model that uses the examination data at the time of the initial examination as input; and the second information processing method includes: a second examination data acquisition step for acquiring examination data at multiple points in time during the hospitalization of the patient to be predicted; and a second prediction step for predicting the subsequent clinical outcome of the patient to be predicted by inputting the examination data acquired in the second examination data acquisition step to a second prediction model that uses the examination data at multiple points in time during hospitalization as input.

[0145] A method for predicting the clinical outcome of a disease according to aspect 12 of the present invention includes a first information processing method for predicting the subsequent clinical outcome of a target patient from examination data at the time of initial examination, and a second information processing method for predicting the subsequent clinical outcome of a target patient from examination data during hospitalization, wherein the first information processing method includes a first training data acquisition step of acquiring training data that includes one or more pairs of examination data at the time of initial examination of a patient suffering from a target disease and the clinical outcome, a first learning step of training a first prediction model that takes the examination data at the time of initial examination as input and predicts the clinical outcome by referring to the training data, a first examination data acquisition step of acquiring examination data at the time of initial examination of the target patient, and inputting the examination data acquired in the first examination data acquisition step to the first prediction model. Therefore, the second information processing method is a method for predicting the clinical outcome of a disease, comprising: a first prediction step of predicting the subsequent clinical outcome of the patient to be predicted; a second training data acquisition step of acquiring training data that includes one or more pairs of examination data during hospitalization of a patient suffering from a target disease and the clinical outcome; a second learning step of training a second prediction model that predicts the clinical outcome by taking examination data at multiple points in time during hospitalization as input and referring to the training data; a second examination data acquisition step of acquiring examination data at multiple points in time during hospitalization of the patient to be predicted; and a second prediction step of predicting the subsequent clinical outcome of the patient to be predicted by inputting the examination data acquired in the second examination data acquisition step to the second prediction model.

[0146] A method for predicting the clinical outcome of a disease according to aspect 13 of the present invention includes: a first information processing method for predicting the subsequent clinical outcome of a patient from examination data at the time of the initial examination; and a second information processing method for predicting the subsequent clinical outcome of a patient from examination data during hospitalization, wherein the first information processing method includes: a first examination data acquisition step for acquiring examination data at the time of the patient's initial examination; and a first prediction step for predicting the subsequent clinical outcome of the patient by inputting the examination data acquired in the examination data acquisition step into a first prediction model that uses the examination data at the initial examination as input; and the second information processing method includes: a second examination data acquisition step for acquiring examination data at multiple points in time during the patient's hospitalization; and a second prediction step for predicting the subsequent clinical outcome of the patient by inputting the examination data acquired in the examination data acquisition step into a second prediction model that uses the examination data at multiple points in time during hospitalization as input.

[0147] A clinical outcome prediction program for a disease according to aspect 14 of the present invention is a program that causes an information processing device to execute a second information processing method for predicting the subsequent clinical outcome of a target patient from hospitalized examination data, wherein the second information processing method includes: a training data acquisition step of acquiring training data that includes one or more pairs of examination data and clinical outcomes of a patient suffering from a target disease during hospitalization; a learning step of training a prediction model that predicts the clinical outcome by taking examination data at multiple points in time during hospitalization as input and referring to the training data; an examination data acquisition step of acquiring examination data at multiple points in time during hospitalization of the target patient; and a prediction step of predicting the subsequent clinical outcome of the target patient by inputting the examination data acquired in the examination data acquisition step to the prediction model.

[0148] A clinical outcome prediction program for a disease according to aspect 15 of the present invention is a program that causes an information processing device to execute a first information processing method for predicting the subsequent clinical outcome of a target patient from initial examination data, wherein the first information processing method includes: a training data acquisition step of acquiring training data that includes one or more pairs of initial examination data of a patient suffering from a target disease and a clinical outcome; a learning step of using the initial examination data as input and training a prediction model that predicts the clinical outcome by referring to the training data; an examination data acquisition step of acquiring initial examination data of the target patient; and a prediction step of predicting the subsequent clinical outcome of the target patient by inputting the examination data acquired in the examination data acquisition step to the prediction model.

[0149] A clinical outcome prediction program for a disease according to aspect 16 of the present invention is a program that causes an information processing device to execute a second information processing method for predicting the subsequent clinical outcome of a patient to be predicted from examination data during hospitalization, wherein the second information processing method includes: an examination data acquisition step of acquiring examination data at multiple points in time during the hospitalization of the patient to be predicted; and a prediction step of inputting the examination data acquired in the examination data acquisition step to a prediction model that uses the examination data at multiple points in time during hospitalization as input to predict the subsequent clinical outcome of the patient to be predicted.

[0150] A clinical outcome prediction program for a disease according to aspect 17 of the present invention is a program that causes an information processing device to execute a first information processing method for predicting the subsequent clinical outcome of a patient to be predicted from examination data at the time of the initial examination, wherein the first information processing method includes an examination data acquisition step for acquiring examination data at the time of the patient to be predicted at the time of the initial examination, and a prediction step for predicting the subsequent clinical outcome of the patient to be predicted by inputting the examination data acquired in the examination data acquisition step to a prediction model that takes the examination data at the time of the initial examination as input.

[0151] The clinical outcome prediction program for a disease according to aspect 18 of the present invention may further include a third program that causes an information processing device to execute a third information processing method, which includes a step of proposing a drug or therapeutic intervention method based on the clinical outcome of the patient predicted by the first information processing method and the second information processing method, in aspect 10 or 11.

[0152] The method for predicting the clinical outcome of a disease according to aspect 18 of the present invention may further include a third information processing method in aspect 12 or 13, which includes the step of proposing a drug or therapeutic intervention method based on the clinical outcome of the patient predicted by the first information processing method and the second information processing method. [Examples]

[0153] [Test design and configuration] This study was conducted as a retrospective cohort study of 382 COVID-19 cases treated at the Department of Respiratory Medicine, Tokyo Shinagawa Hospital, between January 2020 and September 2020. In total, 201 outpatients and 181 inpatients were included as COVID-19 cases. Data from 146 non-COVID-19 cases were also used to verify whether the dynamic mortality risk model is effective for non-COVID-19 cases. Note that the laboratory data and prediction results for severity prediction based on oxygen support levels in the following examples were acquired, predicted, or output by various components of the control unit 11 of the information processing device 1, as an example.

[0154] Data was extracted from electronic medical records, including 84 variables (see Table 1) for prognosis screening at the initial consultation, such as symptoms, background information, and blood / urine test data, and 93 variables (see Table 2) for dynamic risk assessment during hospitalization, such as blood tests and vital signs. Considering that the severity and mortality of COVID-19 have been reported to be influenced by various factors, all variables with less than 15% missing values ​​were used in the analysis. A total of 265 blood / urine tests were performed at the initial consultation, but most were not measured in non-hospitalized patients. We selected tests that were measured in more than half of non-hospitalized patients and made them comparable by severity, resulting in 53 test items and an overall missing value threshold of 15%.

[0155] For the initial prognosis screening, blood and urine test data obtained from 4 days to 1 day after the initial visit were used. For dynamic risk assessment during hospitalization, the most recent study was used until the next study was conducted. Missing values ​​from the initial visit and the dynamic risk assessment were imputed using the missForest package in the statistical analysis software R (Reference 1: Stekhoven DJ, Buhlmann P. MissForest-non-parametric missing value imputation for mixed-type data. Bioinformatics. 2012 Jan 1;28(1):112-8.).

[0156] [Exclusion Criteria and Data Segmentation] One case under 18 years of age and one case with unknown severity changes during hospitalization or unknown clinical outcome were excluded from the analysis. This includes cases where the prognosis was unknown due to transfer to another hospital or termination of observation, and cases where severity changes could not be sufficiently extracted from electronic medical records. Hospitalized patients with a difference of more than 4 days between the initial diagnosis date and the admission date were excluded from prognosis screening but were included in the dynamic risk assessment. The included cases were randomly divided into a training dataset and a validation dataset in a 2:1 ratio, and the distribution of severity outcomes was preserved. The data split was common to both the prediction at the initial consultation and the dynamic risk prediction during hospitalization.

[0157] [Severity prediction] As a predictor of patient severity, we adopted the oxygen support status score proposed by Grein et al. (Reference 2: Grein J, Ohmagari N, Shin D, et al. Compassionate Use of Remdesivir for Patients with Severe Covid-19. N Engl J Med 2020; 382: 2327-36.). The score was an ordinal scale from 1 to 6 based on the type of oxygenation level: 1 for discharge or not hospitalized, 2 for room air, 3 for low-flow oxygen support, 4 for non-invasive intervention including nasal high-flow oxygen therapy and / or non-invasive positive pressure ventilation (NIPPV), 5 for invasive intervention including invasive mechanical ventilation and / or ECMO, and 6 for death. In addition to the initial score of 1 indicating discharge status, outpatients were also given a score of 1.

[0158] [Machine learning model for predicting outcomes at the initial consultation] We predicted the outcome at the initial consultation using a gradient boosting decision tree (GBDT) (Reference 3: Ke G, Meng Q, Finley T, et al. LightGBM: A Highly Efficient Gradient Boosting Decision Tree. In: Guyon I, Luxburg UV, Bengio S, et al., eds. Advances in Neural Information Processing Systems. Curran Associates, Inc., 2017.). GBDT is an advanced version of random forest and utilizes boosting techniques to improve predictive performance. It has the advantage of being able to handle missing data without imputation.

[0159] During the training phase, data imputed by missForest was used, but predictions were made on validation data containing missing values. Among the various implementations of the GBDT algorithm, the Python package LightGBM was used due to its superior computing speed. As a preprocessing step, variables with an absolute value of Spearman correlation coefficient ≥ 0.85 in the training data were excluded, leaving only the variables showing the strongest correlation with severity. Hyperparameter tuning by Bayesian optimization with cross-validation using the Optuna framework and training with the optimal parameters were repeated 100 times. The ensemble mean of the models obtained from the iterations was used as the final prediction model. SHAP (Shapley Additive exPlanations) values ​​were used on the validation data to evaluate the importance of variables in prediction.

[0160] [Random Survival Forest] The risk of death during hospitalization was assessed using a Random Survival Forest (RSF) (Reference 4: Ishwaran H, Kogalur UB, Blackstone EH, Lauer MS. Random survival forests. Ann. Appl. Stat, 2008; 2: 841-60.). The RSF calculates the final hazard function as a collection of hazard functions estimated by the survival forest. Like random forests, the RSF is robust to outliers and allows for accurate risk assessment of event occurrence. The Python package scikit-survival, which includes an RSF implementation, was used for the analysis. To assess the dynamic risk of death, a 7-day cumulative hazard function showing the risk of events occurring within 7 days was calculated for each patient on each day of hospitalization. The variable importance of the RSF model was calculated on the training data using the Permutation Importance function implemented in the Python package eli5. The final predictive model was created by retraining using only the top 10 important variables.

[0161] [ethics] This study was conducted with the approval of the local Institutional Review Boards of RIKEN and Tokyo Shinagawa Hospital (Approval Number 20-A-06). This study was conducted in accordance with the ethical standards of the 1964 Declaration of Helsinki and its subsequent amendments or similar. The requirement to obtain informed consent for participation was waived due to the retrospective nature of this study.

[0162] [result] The median age of the 382 COVID-19 patients was 39 years, with 233 males and 149 females (Table 3). Of these, 201 were outpatients who did not require hospitalization, and the remaining 181 were hospitalized. The median length of hospital stay for hospitalized patients was 11 days. Of the 51 hospitalized patients requiring oxygen support, 30 required low-flow oxygen, 3 required high-flow oxygen, 8 were treated with invasive ventilation / ECMO, and 10 died.

[0163] [Table 3]

[0164] (Screening for the severity and prognosis of COVID-19 at the initial consultation) To examine whether the outcome was already determined at the time of the initial consultation, a prognostic model of COVID-19 severity was constructed based on the information available at the time of the initial consultation (symptoms, background information, blood and urine markers).

[0165] Specifically, using the examination data of the patient to be predicted at the initial consultation as input, the information processing device 1 shown in Figure 1 was used to perform prediction processing, thereby obtaining prediction results for severity prediction based on oxygen support level as proposed by Grein et al. (prediction process). Based on these prediction results, the maximum level of severity that the patient would reach (i.e., severity outcome) was predicted.

[0166] Figure 12 shows the results of the severity prognosis screening at the initial consultation. Data 1201 in Figure 12 is the receiver operating characteristic (ROC) curve and its area under the curve (AUC) for a predictive model of severity outcome based on the level of oxygen support. The 95% confidence interval for the AUC is shown in parentheses. Data 1202 in Figure 12 is a box and jitter plot representing the predicted probability that a validation case will be judged to have a severity exceeding the threshold shown by the dashed line.

[0167] The area under the receiver operating characteristic curve (AUC) was 0.717 for severity ≥2 (hospitalization), 0.878 for ≥3 (oxygen administration), 0.951 for severity outcomes ≥4 (non-invasive ventilation), 0.952 for severity outcomes ≥5 (invasive ventilation), and 0.970 for severity outcomes ≥6 (death) (Data 1201 in Figure 12).

[0168] We examined the agreement rate between predicted probabilities and actual severity outcomes. The results showed that most patients with actual severity levels of 3 or higher had a high probability of being ≥2, but patients with actual severity levels of 1 and 2 showed a wide distribution from low to high predicted probabilities, with no clear distinction between them (Figure 12, top panel of data 1202). For other predictions, there was generally good agreement between predicted probabilities and actual severity, but there were some cases where the severity risk was not adequately determined, such as cases where patients remained mild but had a high probability of becoming severe (Figure 12, data 1202).

[0169] Next, the variable importance of each prediction was evaluated. Figure 13 shows the results of evaluating the variable importance in predicting each severity level of COVID-19 using SHAP values. Data 1301 in Figure 13 shows the variable importance for each prediction model evaluated by SHAP values. The mean of the SHAP absolute values ​​for individual patients in the validation dataset is shown as a bar graph. Data 1302 in Figure 13 is a heatmap showing the symptoms observed at the initial consultation.

[0170] The presence of pneumonia was identified as a top important predictor of whether the severity level would be ≥2 (leftmost panel of data 1301 in Figure 13). Almost all patients with a severity level of 2 or higher requiring hospitalization presented with pneumonia at the initial examination, and all patients with a severity level of 3 or higher presented with pneumonia (data 1302 in Figure 13). Conversely, as the target severity level for prediction increased, the importance of symptoms decreased, but blood test parameters such as lymphocyte count (Lympho-Count), prothrombin time (PT), C-reactive protein (CRP), creatinine (Cre), and amylase (AMY) became top predictors (data 1301 in Figure 13). Body mass index (BMI) was also an important factor in predicting whether the severity level would be ≥4 or ≥5. Age was important for all severity levels.

[0171] Among the important predictors, the distribution of age and blood test results was examined in relation to actual severity prognosis. Figure 14 shows box and jitter plots representing the distribution of important continuous variables by severity group. As shown in Figure 14, patients with severity level 3 or higher had higher age (Age) and BMI, and patients with severity level 4 or higher had higher blood test results for PT, CRP, creatinine (Cre), erythrocyte distribution width (RDW), and blood glucose (GLU). These blood tests, including lymphocyte count (Lympho-Count), estimated glomerular filtration rate (eGFR)-creatinine (eGFR-Cre), and albumin (Alb), were lower in more severe cases, and platelet counts were particularly low in fatal cases. Amylase (AMY) showed a unique distribution, being low in severe patients with severity levels 4, 5, and 6, but very high in some patients with severity level 6.

[0172] (Dynamic mortality risk assessment based on longitudinal data during hospitalization) Outcome screening based on initial consultation information was accurate but incomplete. Therefore, we hypothesized that a patient's prognosis is not fully determined at the initial consultation, and that the prognosis can change depending on the course of treatment after hospitalization. We then examined changes in severity status after hospitalization. Figure 15 shows the changes in severity status during hospitalization for each severity outcome group, illustrating the trend in the number of patients receiving treatment such as ventilation and oxygen administration at predetermined points after hospitalization. As shown in Figure 15, oxygen administration and non-invasive ventilation were initiated within 5 days of hospitalization, while invasive ventilation was often introduced more than 5 days after hospitalization, and death occurred more than 20 days after hospitalization. Furthermore, many deaths occurred without the application of invasive ventilation (top panel of Figure 15). This is because most fatal cases were elderly, and even if their condition worsened, they were not candidates for invasive ventilation or ECMO. These observations suggest that during the weeks from hospitalization to death or discharge, patients undergo changes in their condition that cannot be adequately assessed depending on their oxygen support status.

[0173] Next, we used a Random Survival Forest (RSF) (Reference 4) to assess the risk of death in hospitalized patients. The analysis included 93 variables in total, including background factors such as age and BMI, as well as blood test results and vital signs measured over time during hospitalization. The 7-day cumulative hazard function (CHF), calculated using the RSF as the index, was used to assess the risk of death.

[0174] Specifically, using test data from multiple time points in time, measured sequentially during hospitalization, as input, the information processing device 1 shown in Figure 1 was used to perform prediction processing, thereby obtaining prediction results for severity prediction based on oxygen support level, as proposed by Grein et al. (prediction process). The risk of death was evaluated using a 7-day cumulative hazard function (CHF) calculated using this prediction result as an indicator.

[0175] Figure 16 shows the results of evaluating the mortality risk of COVID-19 patients using the 7-day cumulative hazard function (CHF) for mortality, calculated using RSF as an indicator. As shown in Figure 16, the mortality risk of each patient changed with each day of hospitalization. The mortality risk is shown as the 7-day cumulative hazard function (CHF) calculated using RSF, i.e., the mortality risk within 7 days. The filled circles for fatal cases indicate the last observation day. For all 4 fatal cases in the validation database, an increase in mortality risk was observed approximately one week after hospitalization, and the CHF at the time of death was approximately 0.4. Conversely, in patients who recovered from invasive or non-invasive ventilation, CHF increased approximately one week after hospitalization, similar to the fatal cases, but then decreased, with the maximum CHF never exceeding 0.2. In mild cases that progressed with oxygen administration or room air, there was almost no increase in CHF.

[0176] Figure 17 shows the results of an analysis of variables contributing to RSF prediction. In Figure 17, the importance of important variables that contributed to RSF prediction of mortality was calculated using permutation importance in the training data. As shown in Figure 17, factors related to blood coagulation, such as platelets (Plt) and prothrombin time (PT), were the most important in predicting mortality.

[0177] Figure 18 shows the results confirming the agreement between RSF prediction and the actual clinical course in severe cases. Figure 18 shows the changes in RSF mortality risk and important predictors during hospitalization, using six representative severe cases, including two ECMO-cured cases and four deaths, from the validation data. Oxygen support status is shown between the CHF plot and the plot of important RSF variables. The shaded portion of the graph represents the reference interval.

[0178] As shown in Figure 18, different patients had different combinations of factors associated with changes in CHF. For example, in patient #2, amylase was associated with elevated CHF levels, while in patient #4, elevated PT levels, along with elevated amylase, occurred with elevated CHF levels (Figure 18). Thrombocytopenia was observed in patients #4 and #6, and elevated β-D-glucan levels were observed in patients #3 and #6. Decreased serum calcium levels, increased neutrophil count (Neutro%), and decreased monocyte count (Mono%) were common changes observed in all critically ill patients.

[0179] (Validation of the RSF model for dynamic mortality risk assessment) Regarding the assessment of mortality risk for COVID-19, a machine learning model based on blood test data has been proposed in Nature Machine Intelligence (NMI) 2020 (Non-patent document 3: Yan L, Zhang HT, Goncalves J, et al. An interpretable mortality prediction model for COVID-19 patients. Nature Machine Intelligence 2020; 2: 283-8.). This model (NMI model) was constructed using only samples taken immediately before death or discharge, and lactate dehydrogenase (LDH), lymphocytes, and CRP were identified as key features. Therefore, as a comparative example, mortality risk was assessed based on these features identified by the NMI model.

[0180] Figure 19 shows the results for the comparative example, illustrating the assessment of COVID-19 mortality risk based on LDH, lymphocytes, and CRP identified by the NMI model. Similar to Figure 18, changes in the three predictors and the resulting mortality risk were evaluated for six severe cases, including two ECMO-cured cases and four deaths in the validation data. In the graph in Figure 19, the dashed lines represent the thresholds for the predictors proposed by the NMI model, and the shaded areas represent the reference intervals. The bars at the top of the graph show the mortality prediction based on the NMI model.

[0181] As shown in Figure 19, LDH levels were elevated immediately before death in fatal cases, indicating a high risk. However, even in fatal cases, there were instances where LDH levels did not rise except immediately before death. Furthermore, patient #2 showed elevated LDH levels and was consistently judged to be at high risk of death even as an ECMO survivor.

[0182] In the NMI model (comparative case), the CRP threshold presented was very high (41.2 mg / l), and no cases in this study exceeded this threshold. On the other hand, in severe cases, lymphocyte levels were consistently below the NMI model threshold (14.7%) from the start of hospitalization.

[0183] Figure 20 shows a comparison of the accuracy of the NMI model (comparative example) with that of the RSF model in this embodiment. Figure 20 shows the results of comparing the performance of the RSF model in this embodiment with that of the NMI model (comparative example). For 61 COVID-19 cases in the validation data, the F1 score for distinguishing between recovered and deceased cases is shown for each day until outcome. The bar plot shows the actual number of patients used to evaluate performance, and decreases as the number of days until outcome increases because some patients had not yet been hospitalized.

[0184] As shown in Figure 20, one day before the outcome, both models perfectly predicted mortality with an F1-score of 1.0. However, more than two days before the outcome, the accuracy of the NMI model (comparative example) dropped sharply to an F1-score of approximately 0.5. Conversely, the accuracy of the RSF model in this embodiment did not decline until two weeks before the outcome, suggesting that the RSF model in this embodiment can distinguish between deceased and cured cases well before the outcome.

[0185] Finally, we tested whether the RSF model used to predict mortality was also effective for non-COVID-19 cases. Figure 21 shows the results of examining the effectiveness of the RSF model in predicting mortality and hospital discharge for non-COVID-19 cases. It shows the change in mortality risk for 143 non-COVID-19 cases, including 9 deaths and 134 recoveries. The filled circles for deaths indicate the last observation date.

[0186] As shown in Figure 21, in cases other than COVID-19, CHF levels were elevated in about half of the fatal cases, exceeding 0.2 at the time of death, but not as significantly elevated in the remaining half of the fatal cases. In non-COVID-19 recovered cases, a few showed transiently high CHF levels in the early stages of hospitalization, but most were discharged without their CHF levels exceeding 0.2. Of the 19 non-COVID-19 cases with elevated CHF levels ≥ 0.2, 17 were infections such as bacterial pneumonia, bacterial prostatitis, and urinary tract infections, which were significantly more frequent than other diseases (odds ratio = 14.2, P-value = 0.00070 by Fisher's exact test, Table 4).

[0187] [Table 4]

[0188] [Consideration] In the initial prognosis screening model, different performance and predictors were obtained for each prognostic outcome. Pneumonia was the most important predictor of hospitalization, suggesting that hospitalization is determined by physical examination. Pneumonia has been reported as a risk factor for severe illness, and in this study, all moderate to severe cases requiring oxygen administration had pneumonia at the initial examination (Data 1302 in Figure 13). Although the prognosis model could not adequately distinguish between patients with severity level 1 and those with severity level 2, physical examination is considered a useful screening tool for potential severe cases.

[0189] Among the important factors in predicting oxygen administration, CRP was included in the 4C Mortality Score (Reference 5: Knight SR, Ho A, Pius R, et al. Risk stratification of patients admitted to hospital with covid-19 using the ISARIC WHO Clinical Characterisation Protocol: development and validation of the 4C Mortality Score. BMJ 2020; 370: m3339.). CRP has been reported to be associated with the extent of lung lesions, which supports the importance of CRP in the early identification of moderate to severe cases that may require oxygen administration. However, CRP is not very important in distinguishing between severity outcomes 4, 5, and 6 (Data 1301 in Figure 13). A clear difference in CRP values ​​was observed between severity levels 3 and 4, but there was little difference in severity among severity levels 4 and above (Figure 14), suggesting that inflammation reflected in CRP is an entry point for severity, but the progression of severity is mainly influenced by other factors.

[0190] Creatinine and amylase levels were identified as important predictors of severe cases. In addition to chronic kidney disease, which has been reported as a risk factor for severe illness, COVID-19 has also been reported to cause renal impairment. Amylase is primarily secreted from the pancreas and salivary glands, and previous studies have reported elevated serum amylase levels in severe COVID-19. Elevated serum amylase levels indicate damage to producing tissues or kidney damage related to clearance. Damage to tissues or organs such as the kidneys may be a significant factor in the progression of COVID-19 severity.

[0191] For screening of mortality at initial presentation, RDW, albumin, and platelets were identified as top predictors (Data 1301, Figure 13). RDW has been reported to be associated with COVID-19 mortality. RDW is elevated in cardiovascular disease, cancer, and infection, and has been reported to be associated with increased leukocyte and platelet production followed by slower red blood cell production. Decreased serum albumin levels have also been reported as a predictor of COVID-19 mortality. Coagulation disorders, including thrombocytopenia, are common in COVID-19, and disseminated intravascular coagulation (DIC) has been observed in many COVID-19 deaths. Therefore, these findings suggest abnormalities in hematopoiesis and blood coagulation as early signs of death from COVID-19.

[0192] An information processing device according to one aspect of the present invention can perform a risk assessment that takes into account the time from measurement to event occurrence by machine learning-based survival analysis for predicting COVID-19 mortality, and can accurately estimate the risk of death two weeks before outcome. Several early-changing RSF predictors in fatal cases were identified. Platelet and prothrombin time are the top two early predictors of COVID-19 mortality in the RSF model. As mentioned above, COVID-19 patients are prone to thrombosis in both arteries and veins, and in severe cases, DIC can complicate the patient's condition and lead to death. DIC was observed in only 0.6% of COVID-19 survivors, but occurred very frequently in 71.4% of fatal cases. The results of this embodiment support the idea that coagulation abnormalities that can lead to DIC are specific early predictors of COVID-19 mortality. The predictors in dynamic risk assessment can partially explain the validity of the RSF model for non-COVID-19 cases. Most of the non-COVID-19 cases used for comparison in this embodiment were bacterial respiratory infections, and in severe cases, sepsis and DIC occurred. Therefore, as demonstrated in this embodiment, platelet count and prothrombin time may equally predict the severity and mortality rate of non-COVID-19 infections.

[0193] The two models constructed in this embodiment will likely be useful in various aspects of COVID-19 treatment. The initial prognosis model, which is highly sensitive and screens for potentially severe cases, could be applied to estimate the number of hospital beds and medical devices such as ventilators and ECMO machines needed. Furthermore, due to the recent shortage of hospital beds, the number of patients receiving home care has increased, and there are concerns about rapid exacerbations during home care. It is thought that using the prognosis screening model at the initial consultation can reduce the number of severe cases receiving home care and ensure safer home care. The RSF model for mortality would be useful for the early detection and intervention of patients at risk of death. Coagulation factors, which are important predictors of mortality in the RSF model, are strong targets for early intervention. In COVID-19, heparin administration has been reported to improve the prognosis of severe DIC cases. Using the RSF model, the appropriate timing of anticoagulation therapy can be determined, and the therapeutic effect can be evaluated over time. [Industrial applicability]

[0194] This invention can be used in various aspects of treating any disease, including early detection and intervention of patients at risk of death, and advance preparation of medical resources. Furthermore, this invention is expected to contribute to the development of more efficient treatments and therapeutic strategies to reduce mortality rates in COVID-19 and other infectious diseases, including emerging infectious diseases that may arise in the post-COVID-19 era. [Explanation of Symbols]

[0195] 1,1a,1b,1c,1d Information Processing Device 11,11a,11b,11c,11d Control Unit 111 Data Acquisition Unit 112 Learning Department 113 Prediction Section Output section of 114 115 Specific section 116 Proposal Department 117 Extraction part 1000, 1000a, 1000b, 1000c, 1000d Disease Clinical Outcome Prediction System M1,M2,M3,M4,M5 Information processing method M11 Training Data Acquisition Step M12 Learning Steps M21 Steps to acquire inspection data M22 prediction step M31 Specific Step M32 Proposal Steps M41 Specific Step M42 Extraction Step M51 Prediction Result Acquisition Step M52 Proposal Steps

Claims

1. A first program causes an information processing device to execute a first information processing method that predicts the clinical outcome of a target patient based on examination data from the initial consultation. A second program causes an information processing device to execute a second information processing method that predicts the clinical outcome of a target patient based on test data from during hospitalization. It includes an information processing device that performs the following: The first information processing method is The first training data acquisition step involves obtaining training data that includes multiple pairs of initial examination data and clinical outcomes for patients suffering from the target disease, and A first learning step involves using the initial examination data as input to train a first predictive model that predicts the clinical outcome, by referring to the training data. A first examination data acquisition step involves acquiring examination data from the patient to be predicted at the time of the initial examination, A first prediction step in which the clinical outcome of the patient to be predicted is predicted by inputting the test data acquired in the first test data acquisition step into the first prediction model, Includes, The second information processing method described above is: A second training data acquisition step involves obtaining training data that includes multiple pairs of laboratory test data from patients suffering from the target disease during hospitalization and their clinical outcomes. A second learning step involves using test data from multiple points in time during hospitalization as input to train a second predictive model that predicts the clinical outcome, by referring to the training data. A second test data acquisition step involves acquiring test data at multiple points in time during the hospitalization of the patient to be predicted, A second prediction step involves inputting the test data acquired in the second test data acquisition step into the second prediction model to predict the clinical outcome of the patient to be predicted, Includes, A system for predicting the clinical outcomes of diseases.

2. A first program causes an information processing device to execute a first information processing method that predicts the clinical outcome of a target patient based on examination data from the initial consultation. A second program causes an information processing device to execute a second information processing method that predicts the clinical outcome of a target patient based on test data from during hospitalization. It includes an information processing device that performs the following: The first information processing method is A first examination data acquisition step involves acquiring examination data from the patient to be predicted at the time of the initial examination, A first prediction model that uses initial examination data as input to predict the clinical outcome, wherein the first prediction step predicts the clinical outcome of the target patient by inputting the examination data acquired in the first examination data acquisition step to the first prediction model which has been trained by referring to training data that includes multiple pairs of initial examination data and clinical outcomes of patients suffering from the target disease, Includes, The second information processing method described above is: A second test data acquisition step involves acquiring test data at multiple points in time during the hospitalization of the patient to be predicted, A second predictive model that takes test data from multiple points in time during hospitalization as input to predict the clinical outcome, wherein the second predictive step predicts the clinical outcome of the target patient by inputting the test data acquired in the second test data acquisition step to the second predictive model which has been trained by referring to training data that includes multiple pairs of test data and clinical outcomes of patients suffering from the target disease during hospitalization, Includes, A system for predicting the clinical outcomes of diseases.

3. The clinical outcome prediction system for a disease according to claim 1 or 2, wherein the second prediction model is trained using a random survival forest.

4. The clinical outcome prediction system for a disease according to claim 1 or 2, wherein the first prediction model is trained using a gradient boosting decision tree.

5. A clinical outcome prediction system for a disease according to any one of claims 1 to 4, wherein the information processing device further executes a third program that causes the information processing device to execute a third information processing method which includes the step of proposing a drug or therapeutic intervention method based on the clinical outcome of the patient predicted by the first information processing method and the second information processing method.

6. A first program causes an information processing device to execute a first information processing method that predicts the clinical outcome of a target patient based on examination data from the initial consultation. A second program causes an information processing device to execute a second information processing method that predicts the clinical outcome of a target patient based on test data from during hospitalization. Includes, The first information processing method is The first training data acquisition step involves obtaining training data that includes multiple pairs of initial examination data and clinical outcomes for patients suffering from the target disease, and A first learning step involves using the initial examination data as input to train a first predictive model that predicts the clinical outcome, by referring to the training data. A first examination data acquisition step involves acquiring examination data from the patient to be predicted at the time of the initial examination, A first prediction step in which the clinical outcome of the patient to be predicted is predicted by inputting the test data acquired in the first test data acquisition step into the first prediction model, Includes, The second information processing method described above is: A second training data acquisition step involves obtaining training data that includes multiple pairs of laboratory test data from patients suffering from the target disease during hospitalization and their clinical outcomes. A second learning step involves using test data from multiple points in time during hospitalization as input to train a second predictive model that predicts the clinical outcome, by referring to the training data. A second test data acquisition step involves acquiring test data at multiple points in time during the hospitalization of the patient to be predicted, A second prediction step involves inputting the test data acquired in the second test data acquisition step into the second prediction model to predict the clinical outcome of the patient to be predicted, Includes, A program for predicting clinical outcomes of diseases.

7. A first program causes an information processing device to execute a first information processing method that predicts the clinical outcome of a target patient based on examination data from the initial consultation. A second program causes an information processing device to execute a second information processing method that predicts the clinical outcome of a target patient based on test data from during hospitalization. Includes, The first information processing method is A first examination data acquisition step involves acquiring examination data from the patient to be predicted at the time of the initial examination, A first prediction model that uses initial examination data as input to predict the clinical outcome, wherein the first prediction step predicts the clinical outcome of the target patient by inputting the examination data acquired in the first examination data acquisition step to the first prediction model which has been trained by referring to training data that includes multiple pairs of initial examination data and clinical outcomes of patients suffering from the target disease, Includes, The second information processing method described above is: A second test data acquisition step involves acquiring test data at multiple points in time during the hospitalization of the patient to be predicted, A second predictive model that takes test data from multiple points in time during hospitalization as input to predict the clinical outcome, wherein the second predictive step predicts the clinical outcome of the target patient by inputting the test data acquired in the second test data acquisition step to the second predictive model which has been trained by referring to training data that includes multiple pairs of test data and clinical outcomes of patients suffering from the target disease during hospitalization, Includes, A program for predicting clinical outcomes of diseases.

8. A first information processing method for predicting the clinical outcome of a target patient from initial examination data, A second information processing method for predicting the clinical outcome of a target patient from inpatient test data, Includes, The first information processing method involves one or more processors, The first training data acquisition step involves obtaining training data that includes multiple pairs of initial examination data and clinical outcomes for patients suffering from the target disease, and A first learning step involves using the initial examination data as input to train a first predictive model that predicts the clinical outcome, by referring to the training data. A first examination data acquisition step involves acquiring examination data from the patient to be predicted at the time of the initial examination, A first prediction step in which the clinical outcome of the patient to be predicted is predicted by inputting the test data acquired in the first test data acquisition step into the first prediction model, Includes, The second information processing method involves one or more processors, A second training data acquisition step involves obtaining training data that includes multiple pairs of laboratory test data from patients suffering from the target disease during hospitalization and their clinical outcomes. A second learning step involves using test data from multiple points in time during hospitalization as input to train a second predictive model that predicts the clinical outcome, by referring to the training data. A second test data acquisition step involves acquiring test data at multiple points in time during the hospitalization of the patient to be predicted, A second prediction step involves inputting the test data acquired in the second test data acquisition step into the second prediction model to predict the clinical outcome of the patient to be predicted, Includes, Methods for predicting the clinical outcome of a disease.

9. A first information processing method for predicting the clinical outcome of a target patient from initial examination data, A second information processing method for predicting the clinical outcome of a target patient from inpatient test data, Includes, The first information processing method involves one or more processors, A first examination data acquisition step involves acquiring examination data from the patient to be predicted at the time of the initial examination, A first prediction model that uses initial examination data as input to predict the clinical outcome, wherein the first prediction step predicts the clinical outcome of the target patient by inputting the examination data acquired in the first examination data acquisition step to the first prediction model which has been trained by referring to training data that includes multiple pairs of initial examination data and clinical outcomes of patients suffering from the target disease, Includes, The second information processing method involves one or more processors, A second test data acquisition step involves acquiring test data at multiple points in time during the hospitalization of the patient to be predicted, A second predictive model that takes test data from multiple points in time during hospitalization as input to predict the clinical outcome, wherein the second predictive step predicts the clinical outcome of the target patient by inputting the test data acquired in the second test data acquisition step to the second predictive model which has been trained by referring to training data that includes multiple pairs of test data and clinical outcomes of patients suffering from the target disease during hospitalization, Includes, Methods for predicting the clinical outcome of a disease.

10. A program that causes an information processing device to execute a second information processing method for predicting the clinical outcome of a target patient from test data obtained during hospitalization, The second information processing method described above is: A training data acquisition step involves obtaining training data that includes multiple pairs of laboratory test data from patients suffering from the target disease during hospitalization and their clinical outcomes. A learning step involves using test data from multiple points in time during hospitalization as input to train a predictive model that predicts the clinical outcome, by referring to the training data, A step of acquiring test data to acquire test data at multiple points in time during the hospitalization of the patient to be predicted, A prediction step in which the clinical outcome of the patient to be predicted is predicted by inputting the test data acquired in the test data acquisition step into the prediction model, A disease clinical outcome prediction program that includes [specific features / features].

11. A program that causes an information processing device to execute a first information processing method for predicting the clinical outcome of a target patient based on examination data from the initial consultation, The first information processing method is A training data acquisition step involves obtaining training data that includes multiple pairs of initial examination data and clinical outcomes for patients suffering from the target disease, and A learning step in which a predictive model that predicts the clinical outcome is trained by taking the initial examination data as input and referring to the training data, A step of acquiring examination data to acquire examination data at the initial examination of the patient to be predicted, A prediction step in which the clinical outcome of the patient to be predicted is predicted by inputting the test data acquired in the test data acquisition step into the prediction model, A disease clinical outcome prediction program that includes [specific features / features].

12. A program that causes an information processing device to execute a second information processing method for predicting the clinical outcome of a target patient from test data obtained during hospitalization, The second information processing method described above is: A step of acquiring test data to acquire test data at multiple points in time during the hospitalization of the patient to be predicted, A predictive model that takes test data from multiple points in time during hospitalization as input to predict the clinical outcome, wherein the predictive model is trained by referring to training data that includes multiple pairs of test data and clinical outcomes of patients suffering from the target disease during hospitalization, and the test data acquired in the test data acquisition step is input to predict the clinical outcome of the patient to be predicted. A disease clinical outcome prediction program that includes [specific features / features].

13. A program that causes an information processing device to execute a first information processing method for predicting the clinical outcome of a target patient based on examination data from the initial consultation, The first information processing method is A step of acquiring examination data to acquire examination data at the initial examination of the patient to be predicted, A prediction model that uses initial examination data as input to predict the clinical outcome, wherein the prediction step involves inputting the examination data acquired in the examination data acquisition step to a prediction model that has been trained by referring to training data that includes multiple pairs of initial examination data and clinical outcomes of patients suffering from the target disease, thereby predicting the clinical outcome of the patient to be predicted. A disease clinical outcome prediction program that includes [specific features / features].

14. A disease clinical outcome prediction program according to claim 6 or 7, further comprising a third program that causes an information processing device to execute a third information processing method, which includes the step of proposing a drug or therapeutic intervention method based on the clinical outcome of the patient predicted by the first information processing method and the second information processing method.

15. A method for predicting the clinical outcome of a disease according to claim 8 or 9, further comprising a third information processing method which includes the step of proposing a drug or therapeutic intervention method based on the clinical outcome of the patient predicted by the first information processing method and the second information processing method.

16. A method for predicting the clinical outcome of a patient to be predicted from examination data during hospitalization, The above method involves one or more processors, A training data acquisition step involves obtaining training data that includes multiple pairs of laboratory test data from patients suffering from the target disease during hospitalization and their clinical outcomes. A learning step involves using test data from multiple points in time during hospitalization as input to train a predictive model that predicts the clinical outcome, by referring to the training data, A step of acquiring test data to acquire test data at multiple points in time during the hospitalization of the patient to be predicted, A prediction step in which the clinical outcome of the patient to be predicted is predicted by inputting the test data acquired in the test data acquisition step into the prediction model, A method for predicting the clinical outcome of a disease, which includes [the following].

17. A method for predicting the clinical outcome of a patient to be predicted from examination data at the initial consultation, The above method involves one or more processors, A training data acquisition step involves obtaining training data that includes multiple pairs of initial examination data and clinical outcomes for patients suffering from the target disease, and A learning step in which a predictive model that predicts the clinical outcome is trained by taking the initial examination data as input and referring to the training data, A step of acquiring examination data to acquire examination data at the initial examination of the patient to be predicted, A prediction step in which the clinical outcome of the patient to be predicted is predicted by inputting the test data acquired in the test data acquisition step into the prediction model, A method for predicting the clinical outcome of a disease, which includes [the following].

18. A method for predicting the clinical outcome of a patient to be predicted from examination data during hospitalization, The aforementioned method, A step of acquiring test data to acquire test data at multiple points in time during the hospitalization of the patient to be predicted, A predictive model that takes test data from multiple points in time during hospitalization as input to predict the clinical outcome, wherein the predictive model is trained by referring to training data that includes multiple pairs of test data and clinical outcomes of patients suffering from the target disease during hospitalization, and the test data acquired in the test data acquisition step is input to predict the clinical outcome of the patient to be predicted. A method for predicting the clinical outcome of a disease, which includes [the following].

19. A method for predicting the clinical outcome of a patient to be predicted from examination data at the initial consultation, The aforementioned method, A step of acquiring examination data to acquire examination data at the initial examination of the patient to be predicted, A prediction model that uses initial examination data as input to predict the clinical outcome, wherein the prediction step involves inputting the examination data acquired in the examination data acquisition step to a prediction model that has been trained by referring to training data that includes multiple pairs of initial examination data and clinical outcomes of patients suffering from the target disease, thereby predicting the clinical outcome of the patient to be predicted. A method for predicting the clinical outcome of a disease, which includes [the following].

Images