Medical support device, medical support method, and medical support program

Abstract

To provide a medical support device, method and program, which enable highly accurate fall prediction while reducing the burden on patients and medical institutions.SOLUTION: A medical support method disclosed herein involves causing a processor to: acquire medical information including date information associated with medical procedures performed on a target patient and predicted date information for predicting a fall for the target patient, derive fall prediction information of the target patient using a prediction model machine-trained to predict a fall of the target patient based on the number of elapsed days derived from the date information and the predicted date information, and notify the target patient of the fall prediction information.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The present disclosure relates to a medical support device, method, and program. [Background technology]

[0002] Falls, which can be caused by muscle weakness due to aging or medication, are a serious incident that strongly affects a patient's quality of life (QOL) and prognosis. Therefore, various methods for predicting falls have been proposed to prevent patient falls. For example, Patent Document 1 proposes a method for predicting a patient's fall based on information acquired by sensors attached to the patient and their footwear. Furthermore, Patent Document 2 proposes a method for predicting a fall using a learning model trained to predict falls based on recorded information including the patient's activity information, vital signs, planning information such as a treatment plan, and items recorded in the patient's medical record, and notifying medical professionals of the fall. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Special Publication No. 2006-228024 [Patent Document 2] Japanese Patent Publication No. 2022-169193 Summary of the Invention [Problem to be solved by the invention]

[0004] The method described in Patent Document 1 requires patients to wear a wearable device equipped with a sensor. Medical institutions are also required to manage the operation and inventory of the wearable device. Therefore, the method described in Patent Document 1 places a heavy burden on both patients and medical institutions. Meanwhile, the method described in Patent Document 2 can predict falls based on various information, but does not consider the number of days elapsed between the date and time of prescription of medication or treatment and the timing of predicting a fall, resulting in low accuracy in fall prediction.

[0005] The present disclosure has been made in consideration of the above circumstances, and aims to enable fall prediction to be performed with high accuracy while reducing the burden on patients and medical institutions. [Means for solving the problem]

[0006] A medical support device according to the present disclosure includes at least one processor, The processor Acquire medical information including date information associated with medical procedures for the target patient and predicted date information for predicting a fall for the target patient; Deriving fall prediction information for the target patient using a prediction model trained by machine learning to predict a fall for the target patient based on the number of days elapsed derived from the date information and the predicted date information; Notify fall prediction information.

[0007] The medical support method according to the present disclosure includes acquiring medical information including date information associated with a medical procedure for a target patient and predicted date information for predicting a fall of the target patient; Deriving fall prediction information for the target patient using a prediction model trained by machine learning to predict a fall for the target patient based on the number of days elapsed derived from the date information and the predicted date information; Notify fall prediction information.

[0008] The medical support program according to the present disclosure includes a procedure for acquiring medical information including date information associated with a medical procedure for a target patient and predicted date information for predicting a fall of the target patient; A step of deriving fall prediction information for a target patient using a prediction model that has been machine-learned to predict a fall for a target patient based on the number of days elapsed derived from the date information and the predicted date information; The computer is caused to execute a procedure for notifying the user of the fall prediction information. [Effects of the Invention]

[0009] According to the present disclosure, it is possible to predict falls with high accuracy while reducing the burden on patients and medical institutions. [Brief explanation of the drawings]

[0010] [Figure 1] FIG. 1 is a diagram showing a schematic configuration of a medical support system to which a medical support device according to an embodiment of the present invention is applied. [Figure 2] FIG. 1 is a diagram showing the hardware configuration of a medical support device according to an embodiment of the present invention. [Figure 3] Functional configuration diagram of a medical support device according to this embodiment [Figure 4] Diagram showing the training data used to train a predictive model [Figure 5] Diagram showing a list of high-risk patients [Figure 6] Diagram showing displayed contributions [Figure 7] A diagram showing the message contained in the notification [Figure 8] Diagram showing a list of scheduled patients [Figure 9] A flowchart showing the processing performed in this embodiment DETAILED DESCRIPTION OF THE INVENTION

[0011] Hereinafter, an embodiment of the present disclosure will be described with reference to the drawings. First, the configuration of a medical support system to which a medical support device according to this embodiment is applied will be described. FIG. 1 is a diagram showing a schematic configuration of the medical support system. In the medical support system 10 shown in FIG. 1, a medical support device 1, a management server 2, and multiple client terminals 3 are connected to each other via a network 4 so that they can communicate with each other. The medical support system 10 is a system used for at least one facility (e.g., a hospital) that handles multiple pieces of medical information.

[0012] The medical support device 1 predicts a fall of the patient as described below, based on information acquired from the management server 2. The detailed configuration of the medical support device 1 will be described later.

[0013] The management server 2 is composed of a server computer or the like that manages various information related to multiple patients. The management server 2 may be a cloud server. The management server 2 manages electronic medical records of patients, medical images acquired by photographing patients, documents created by medical professionals, etc. The medical images are medical images of patients acquired by various modalities. Examples of documents include reports created by doctors, technicians, pharmacists, nurses, etc., or reports created by modalities that acquired medical images of patients.

[0014] The management server 2 acquires basic patient information, disease information, medication information, surgery information, and examination information from electronic medical records, medical images, and documents for multiple patients, and manages the acquired basic patient information, disease information, medication information, surgery information, and examination information by creating a database. In this embodiment, the management server 2 also acquires keyword information by referring to the electronic medical records, and manages the keyword information by creating a database including medical information.

[0015] Basic patient information includes age, sex, blood type, height, weight, BMI, medical departments visited and the number of visits, number of visits, average time of visit, most recent medical department, number of hospitalizations, most recent date of hospitalization, length of hospitalization, and most recent date of discharge.

[0016] The disease information includes the number of times a fall-related disease name has been assigned, the name of the disease in ICD10 code units or in the major classification units of the ICD10 code (e.g., M for musculoskeletal system), and the date on which the disease name was assigned (diagnosis date). The disease information is obtained for both confirmed disease names assigned after a definite diagnosis by a doctor and suspected disease names assigned when the possibility of a disease is suspected based on the patient's symptoms or test results, etc. ICD10 is the 10th revision of the International Statistical Classification of Diseases and Related Health Problems, a medical classification list created by the World Health Organization.

[0017] The medication information includes the number of times a medication related to falls has been prescribed (e.g., "anti-anxiety medication" three times), the prescription date of the medication related to falls (e.g., anti-anxiety medication), the number of consecutive days if the medication was administered, and the total number of medications and types prescribed up to the reference date. The medication information also includes side effect information indicating whether or not each medication has a side effect. For medications related to falls, medications with side effect information related to falls (e.g., "syncope," "dizziness," "nausea," etc.) may be derived based on the side effect information. Injections are treated in the same way as medication prescriptions. The prescription date may include the time in addition to the date.

[0018] The test information includes whether or not a specimen test item was tested, the test date for the specimen test item, the number of specimen tests performed by each department, the number of times the test value exceeded or fell below the test reference value, the total or average number of test items, and the total or average number of types of test items. In addition to specimen tests, similar information is also included regarding the test results of physiological tests, endoscopic tests, radiological tests, etc. The test date may include the time in addition to the date.

[0019] The surgery information includes whether or not the patient has had a previous surgery (number of times), the date of surgery, the surgical method, etc. The date of surgery may include the time in addition to the date.

[0020] Keyword information includes the number of medical record entries written up to the reference date, the presence or absence of keywords related to falls (e.g., "fall," "collapse," "dizziness," "unsteadiness," "cane," and "walker"), and the total number of times the keywords appeared.

[0021] In response to a request from the medical support device 1 (to be described later), the management server 2 acquires medical information from the database and provides it to the medical support device 1 via the network 4.

[0022] The client terminal 3 is a terminal device owned by medical professionals such as doctors, engineers, pharmacists, nurses, and other hospital staff. Examples of the client terminal 3 include workstations, personal computers, tablet terminals, smartphones, and smartwatches.

[0023] A WAN (Wide Area Network) is an example of the network 4. Note that a WAN is merely an example, and the network 4 may be configured as at least one of a LAN (Local Area Network) and a WAN.

[0024] In addition, instead of the management server 2, the medical support system 10 may have a separate electronic medical record server that manages electronic medical records, an image management server that manages images, and a document management server that manages documents, which are connected to each other via the network 4 so that they can communicate with each other.

[0025] 1, the client terminal 3 is connected to the medical support device 1 via a single network 4, but the technology of the present disclosure is not limited to this. For example, the client terminal 3 may be connected to the medical support device 1 via a network different from the network 4 to which the management server 2 is connected.

[0026] The medical support program of this embodiment is installed in the medical support device 1. The medical support device 1 may be a workstation or personal computer installed in a hospital, or may be a server computer. The medical support program is stored in an externally accessible state in a storage device of another server computer connected to a network or in network storage (neither of which is shown), and is downloaded and installed in the medical support device 1 upon request. Alternatively, the program may be recorded on a recording medium such as a DVD (Digital Versatile Disc) or CD-ROM (Compact Disc Read Only Memory) and distributed, and then installed in the medical support device 1 from the recording medium.

[0027] Fig. 2 is a diagram showing the hardware configuration of a medical support device according to this embodiment. As shown in Fig. 2, the medical support device 1 includes a CPU (Central Processing Unit) 11, a display 14, an input device 15, a memory 16, and a network I / F (Interface) 17 connected to a network 4. The CPU 11, the display 14, the input device 15, the memory 16, and the network I / F 17 are connected to a bus 19. The CPU 11 is an example of a processor in the present disclosure.

[0028] The memory 16 includes the storage unit 13 and a RAM (Random Access Memory) 18. The RAM 18 is a memory for primary storage, and is, for example, a RAM such as an SRAM (Static Random Access Memory) or a DRAM (Dynamic Random Access Memory).

[0029] The storage unit 13 is a non-volatile memory, and is realized by, for example, at least one of a hard disk drive (HDD), a solid state drive (SSD), an electrically erasable programmable read-only memory (EEPROM), and a flash memory. The storage unit 13, which serves as a storage medium, stores the medical support program 12 according to this embodiment. The CPU 11 reads the medical support program 12 from the storage unit 13, expands it in the RAM 18, and executes the expanded medical support program 12.

[0030] The display 14 is a device for displaying various screens, such as a liquid crystal display or an EL (Electro Luminescence) display. The input device 15 is a device for a user to input, such as at least one of a keyboard, a mouse, a microphone for voice input, a touchpad for proximity input including contact, and a camera for gesture input. The network I / F 17 is an interface for connecting to the network 4.

[0031] Next, the functional configuration of the medical support device according to this embodiment will be described. Fig. 3 is a diagram showing the functional configuration of the medical support device according to this embodiment. As shown in Fig. 3, the medical support device 1 includes an information acquisition unit 21, a prediction unit 22, and a notification unit 23. When the CPU 11 executes the medical support program 12, the CPU 11 functions as the information acquisition unit 21, the prediction unit 22, and the notification unit 23.

[0032] The information acquisition unit 21 acquires, at predetermined times, medical information including date information associated with medical procedures for a target patient who is predicted to fall, and predicted date information predicting a fall for the target patient, from the management server 2. Here, the management server 2 manages medical information for multiple patients, but the information acquisition unit 21 acquires medical information for a target patient who is associated with predicted date information as a planned hospital visit date. Therefore, the number of target patients is smaller than the number of patients managed by the management server 2.

[0033] The predicted date information is information that indicates the date on which a fall, described below, is predicted for the target patient, and may be, for example, the next scheduled visit date of the target patient. In addition to the date, the predicted date information may also include time information, such as the scheduled time of the visit. Furthermore, the predicted date information is not limited to a single date, but may also include multiple dates. Note that if there are multiple dates, the multiple dates may be consecutive dates specified by a start date and an end date, or non-consecutive dates such as all Tuesdays of the following month.

[0034] In this embodiment, the information acquisition unit 21 acquires medical information about all target patients who have an appointment to visit the hospital on the day, i.e., whose predicted date information is for the day, for example, before the reception hours on the day (for example, the night before). Note that the information acquisition unit 21 may acquire medical information and predicted date information about a target patient when an appointment for the next visit is made for the target patient.

[0035] Examples of medical procedures performed on patients include at least one of the following: prescribing medication, conducting tests, performing surgery, hospitalization / discharge, and performing a diagnosis. The target of a medical procedure may be a medical procedure that may affect falls. The determination of whether or not a medical procedure may affect falls may be based on, for example, the presence or absence of a specific description of dizziness or other side effects in the medical record, or may be specified by a medical professional. In this embodiment, the date information associated with a medical procedure is at least one of the prescription date of medication included in the medication information of the medical information, the examination date included in the examination information, the surgery date included in the surgery information, the admission date and discharge date included in the patient basic information, and the diagnosis date included in the disease information.

[0036] In the case where the drug information includes prescription dates for multiple types of drugs, the test information includes test dates for multiple types of tests, the surgery information includes multiple types of surgery dates, or the patient basic information includes multiple admission dates, discharge dates, and diagnosis dates, in this embodiment, at least one of the most recent prescription date, most recent test date, most recent surgery date, most recent admission date, most recent discharge date, and most recent diagnosis date is used as date information for the predicted date information. Note that the drug prescription date may include the prescription start date and prescription end date. In this case, the prescription end date can be used as date information. If the prescription date and examination date include a time, the date information also includes time information.

[0037] If the patient's medical information includes all of the prescription date, examination date, surgery date, hospitalization date, discharge date, and diagnosis date, the date closest to the predicted date information can be used as the date information. Alternatively, one or more predetermined dates from the prescription date, examination date, surgery date, hospitalization date, discharge date, and diagnosis date can be used. The most recent prescription date, examination date, surgery date, hospitalization date, discharge date, and diagnosis date are examples of specific date information disclosed herein.

[0038] The prediction unit 22 derives fall prediction information for a target patient using a prediction model 24 that has been machine-learned to predict a fall for the target patient based on the number of elapsed days based on date information and predicted date information. The prediction model 24 is stored in the storage unit 13. In this embodiment, the prediction unit 22 derives the number of elapsed days from the date based on the date information to the predicted date based on the predicted date information from the date information and the predicted date information. For example, the prediction unit 22 derives the number of elapsed days from at least one of the prescription date included in the drug information, the examination date included in the examination information, and the surgery date included in the surgery information to the predicted date based on the predicted date information. Note that when multiple date information from the prescription date, examination date, and surgery date is used, the prediction unit 22 derives multiple elapsed days.

[0039] The prediction unit 22 may compare the number of elapsed days with a predetermined threshold value Th1 and derive fall prediction information only when the number of elapsed days is equal to or less than the threshold value Th1. The threshold value Th1 may be, for example, 10 days, but is not limited to this. A different threshold value Th1 may be used depending on the classification of medical information associated with the date information. That is, a different threshold value Th1 may be used depending on whether the date based on the date information is the prescription date included in the drug information, the examination date included in the examination information, the surgery date included in the surgery information, the admission date or discharge date included in the patient basic information, or the diagnosis date included in the disease information. Furthermore, if there are multiple drugs, a different threshold value Th1 may be used for each prescription date of the drug. The threshold value Th1 is an example of a first threshold value in the present disclosure.

[0040] The prediction model 24 receives input of medical information and the number of days elapsed. If multiple numbers of days elapsed are derived, multiple numbers of days elapsed are input. Specifically, a feature vector representing the medical information and the number of days elapsed is input. The feature vector is a vector whose elements are multiple pieces of information included in the medical information and the number of days elapsed. The prediction model 24 then outputs fall prediction information based on the input medical information and the number of days elapsed. The prediction model 24 derives the fall prediction information based on the medical information and the number of days elapsed.

[0041] For this purpose, the prediction model 24 is constructed by machine learning a learning model using training data for patients who have actually fallen, such as medical information, the number of days since the medical procedure was performed, and information on whether or not the patient has fallen. In this embodiment, the training data is acquired from a database managed by the management server 2.

[0042] A decision tree model, for example, can be used as a learning model for constructing the predictive model 24. The decision tree model may be a LightGBM (Light Gradient Boosting Machine) model based on a gradient boosting algorithm. LightGBM is a free and open-source distributed gradient boosting framework for machine learning. LightGBM is based on a decision tree algorithm and is used for ranking, classification, and other machine learning tasks.

[0043] FIG. 4 is a diagram showing training data used to train the prediction model 24. As shown in FIG. 4, training data 30 includes the number of days 31 elapsed since the date of medical treatment, medical information 32, and information 33 on whether or not a fall occurred. In the training data 30 shown in FIG. 4, the number of days 31 elapsed since the date of prescription included in the medication information is the number of days elapsed since the date of prescription, e.g., 7 days. The medical information 32 is obtained for the actual patient used in training data 30 and includes specific information such as basic patient information, disease information, medication information, surgery information, examination information, and keyword information. The information 33 on whether or not a fall occurred is "yes." Note that multiple elapsed days may be derived, in which case the number of elapsed days 31 in training data 30 also includes multiple days.

[0044] During learning, a feature vector representing the number of days elapsed 31 and medical information 32 is derived from training data 30, and the derived feature vector is input to the learning model. The feature vector has dimensions according to the number of days elapsed 31 and the number of pieces of medical information 32. The learning model is then caused to output the predicted probability of a fall as a value between 0 and 1, for example. The output value is then compared with information 33 about the presence or absence of a fall contained in training data 30, and the difference is derived as a loss.

[0045] The correct data is 1 if a fall has occurred and 0 if no fall has occurred. The prediction model 24 is constructed by repeating learning until the loss falls below a predetermined threshold or until a predetermined number of learning rounds have been completed. The prediction model 24 constructed in this manner outputs the predicted probability of a fall as a value between 0 and 1 when the number of days elapsed since the medical procedure and medical information are input. Generally, the fewer days elapsed since the medical procedure, the higher the possibility of a fall. Therefore, the smaller the number of days elapsed input into the prediction model 24, the higher the predicted probability of a fall output by the prediction model 24. Alternatively, the prediction model 24 may be constructed to output the presence or absence of a fall as a value of 1 or 0.

[0046] In this embodiment, when the prediction unit 22 predicts whether or not a fall will occur, it may apply SHAP (Shapley Additive exPlanations), a type of explainable AI technology, to the prediction model 24 to derive the contribution of each element (i.e., feature) of the feature vector input to the prediction model 24. SHAP is a method for determining the contribution of each variable (feature) to the model's prediction results and is based on a concept called the Shapley value. The Shapley value was originally proposed in a field called cooperative game theory. In cooperative game theory, in a game in which multiple players cooperate to complete a game to earn a reward, the main task is to determine how to fairly distribute the reward according to each player's contribution. In machine learning, a predicted value is calculated by combining multiple types of feature values. Therefore, the contribution of the feature value to the predicted value can be derived by replacing the feature value with the player and the reward with the predicted value.

[0047] The notification unit 23 notifies the fall prediction information derived by the prediction unit 22. The notification is sent via the network 4 to a client terminal 3 of a predetermined medical professional, the target patient, or the target patient's family. For example, if the fall prediction information is derived before the reception hours of the day, the notification unit 23 identifies a target patient who is highly likely to fall or for whom prediction information indicating a fall has been obtained (hereinafter referred to as a high-risk patient). A high-risk patient is a target patient for whom the predicted probability of a fall output by the prediction model 24 is equal to or greater than a threshold value Th2, or a target patient for whom the prediction model 24 has output a prediction result indicating a fall. The threshold value Th2 is an example of a second threshold value in the present disclosure.

[0048] Furthermore, instead of or in addition to target patients whose predicted probability of fall is equal to or greater than threshold value Th2, notifier 23 may select target patients in descending order of predicted probability of transfer so that the number of target patients is equal to or less than threshold value Th3. The second threshold value Th2 and the third threshold value Th3 may be predetermined set values ​​or variable values ​​that vary according to user input, etc. For example, at least one of second threshold value Th2 and third threshold value Th3 may be set for each facility or department.

[0049] The notification unit 23 creates a list of identified high-risk patients and notifies the fall prediction information by sending the created list to the client terminal 3 of the medical professional associated with the target patient. Examples of the medical professional associated with the target patient include the target patient's attending physician, outpatient nurses, reception staff, or staff at the hospital entrance.

[0050] Figure 5 is a diagram showing a list of high-risk patients. As shown in Figure 5, the list 40 of high-risk patients includes the high-risk patient's name 41, patient number 42, and appointment date and time 43. By referring to the list 40, medical staff can take measures to prevent patients from falling, such as stationing staff at the hospital entrance to wait when a high-risk patient arrives.

[0051] It should be noted that if the contribution rate is derived by the prediction model 24, the list 40 includes a contribution rate button 44. When the contribution rate button 44 is selected, the contribution rate of the feature values ​​input to the prediction model 24, i.e., the medical information of the target patient and the number of days elapsed, to the fall prediction information, is displayed on the client terminal 3. It should be noted that instead of displaying the contribution rate button 44, the contribution rate may be displayed in advance on the list 40.

[0052] Fig. 6 is a diagram showing the displayed contribution degrees. As shown in Fig. 6, the contribution degree 47 displays, in descending order of contribution degree, the number of times a drug has been prescribed, the number of times a disease has been given, age, the number of days since the drug was prescribed, the number of days since the examination, weight, and height. In this way, by notifying the contribution degree, it is possible to know which feature of the medical information or the number of days since the examination contributes to the fall prediction information.

[0053] The notification unit 23 may also notify the high-risk patient or their family member of their high-risk patient's terminal device that they are at high risk of falling. FIG. 7 is a diagram showing an example of the notification. As shown in FIG. 7, the notification 48 includes the message, "Today's appointment at Hospital A is at 10:00 AM. Please be careful as there is a high possibility of falling." The notification may be sent by email, a messaging app, or telephone. This allows the target patient or their family member to pay more attention to the target patient's movements and take measures to prevent the target patient from falling.

[0054] Furthermore, if fall prediction information is acquired at the time a patient's next visit is scheduled, a notification is sent to the healthcare professional associated with that patient. The notification is sent by adding an alert to the list of scheduled patients displayed on the healthcare professional's client terminal 3. FIG. 8 shows another example of the notification. As shown in FIG. 8, the scheduled patient list 50 includes the name 51 of the patient who has scheduled a visit, the patient number 52, the appointment date and time 53, and the fall risk 54. In FIG. 8, for high-risk patients, a mark 55 indicating a high fall risk is added to the fall risk 54. This allows healthcare professionals to easily recognize which scheduled patients have a high fall risk.

[0055] If there is no high-risk patient in the fall prediction information derived by the prediction unit 22, the notification unit 23 may notify that there is no high-risk patient.

[0056] Next, the processing performed in this embodiment will be described. Fig. 9 is a flowchart showing the processing performed in this embodiment. The processing starts at a predetermined timing, and the information acquisition unit 21 acquires medical information including date information associated with medical treatment for a patient and predicted date information for predicting a patient fall from the management server 2 (information acquisition; step ST1). Next, the prediction unit 22 derives fall prediction information for the patient using a prediction model 24 that has been machine-learned to predict a patient fall (step ST2). Then, the notification unit 23 notifies the fall prediction information (step ST3), and the processing ends.

[0057] In this manner, in this embodiment, fall prediction information for a patient is derived based on medical information, including date information associated with a medical procedure, and prediction date information. This eliminates the need for patients to wear a wearable device equipped with a sensor, and eliminates the need for doctors to manage the operation and inventory of wearable devices. Fall prediction is also possible, taking into account the number of days elapsed between the medical procedure and the time the fall is predicted. Therefore, this embodiment enables highly accurate fall prediction while reducing the burden on patients and medical institutions.

[0058] It is known that the likelihood of a fall increases when a patient receives a fall-related medication consecutively or multiple times. For example, the long-term administration of steroids increases the risk of osteoporosis, so attention must be paid to the possibility of a fall. For this reason, fall prediction information may be derived that is weighted according to the number of consecutive days of medical procedures, such as the number of consecutive days of medication administration, or the number of times a medical procedure has been performed, such as the number of times a medication has been administered within a predetermined period from the date indicated by the appointment date information.

[0059] In this case, the prediction unit 22 may weight the predicted probability of a fall output by the prediction model 24, or may weight the number of elapsed days input to the prediction model 24. When weighting the predicted probability output by the prediction model 24, the fall prediction information may be derived by weighting the predicted probability so that the greater the number of consecutive days or the number of medical procedures. Specifically, the fall prediction information may be derived by multiplying the predicted probability by weighting factors of 3.0, 2.0, and 1.0 for 30, 20, and 10 consecutive days, respectively. For example, if the number of consecutive days of administration of a medication related to falls is 30 days and the predicted probability output by the prediction model 24 is 0.125, the prediction unit 22 may derive the fall prediction information by calculating 0.125 × 3.0 = 0.375. Regarding the number of times of medical procedures, for example, fall prediction information may be derived by multiplying the predicted probability by weighting factors of 3.0, 2.0, and 1.0 for 10, 5, and 3 times, respectively.

[0060] Furthermore, when weighting the number of elapsed days in the prediction model 24, the fall prediction information can be derived by weighting the number of elapsed days so that the greater the number of consecutive days or the number of medical procedures, the smaller the number of elapsed days input into the prediction model 24. Specifically, with regard to the number of consecutive days, the fall prediction information can be derived by multiplying the number of elapsed days by weighting factors of 0.7, 0.8, and 0.9 for 30, 20, and 10 days, respectively. With regard to the number of medical procedures, for example, the fall prediction information can be derived by multiplying the prediction probability by weighting factors of 0.7, 0.8, and 0.9 for 10, 5, and 3 times, respectively.

[0061] Here, the smaller the number of elapsed days, the smaller the prediction probability output by the prediction model 24. Therefore, by weighting the number of elapsed days so that the greater the number of consecutive days or frequency, the smaller the number of elapsed days input to the prediction model 24, it is possible to derive fall prediction information according to the risk of fall, such as continuous or multiple administration of a drug.

[0062] On the other hand, instead of weighting the predicted probability output by the prediction model 24 or the number of elapsed days input into the prediction model 24, the prediction model 24 may be constructed so that the predicted probability of a fall is derived when the number of consecutive days or the number of medical procedures is input in addition to the number of elapsed days and medical information.

[0063] In the above embodiment, the fall prediction information is derived only when the number of elapsed days is equal to or less than the threshold value Th1, but this is not limited to this. The fall prediction information may be derived for all elapsed days.

[0064] In the above embodiment, the expected date of hospital visit is used as the predicted date information, but this is not limited to this. Information on any date specified by the operator can be used as the predicted date information.

[0065] Furthermore, although the contribution degree is derived in the above embodiment, the present invention is not limited to this, and the derivation and notification of the contribution degree may be omitted.

[0066] In the above embodiment, the fall prediction information is derived by inputting all of the basic patient information, disease information, medication information, surgery information, examination information, and keyword information as medical information into the prediction model 24, but this is not limited to this. Only a portion of the basic patient information, disease information, medication information, surgery information, examination information, and keyword information may be input into the prediction model 24 to derive the fall prediction information.

[0067] Furthermore, in the above embodiment, the following various processors can be used as the hardware structure of the processing units that perform various processes, such as the information acquisition unit 21, the prediction unit 22, and the notification unit 23. As described above, the various processors include a CPU, which is a general-purpose processor that executes software (programs) and functions as various processing units, as well as dedicated electrical circuits that are processors having a circuit configuration specifically designed to perform specific processes, such as a GPU (Graphics Processing Unit), a Programmable Logic Device (PLD), which is a processor whose circuit configuration can be changed after manufacture, such as an FPGA (Field Programmable Gate Array), and an ASIC (Application Specific Integrated Circuit).

[0068] A single processing unit may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs or a combination of a CPU and an FPGA). Also, multiple processing units may be configured with a single processor.

[0069] Examples of configuring multiple processing units with a single processor include, first, a form in which one processor is configured with a combination of one or more CPUs and software, and this processor functions as multiple processing units, as typified by computers such as client and server. Second, a form in which a processor is used to realize the functions of an entire system including multiple processing units with a single IC (Integrated Circuit) chip, as typified by systems on chips (SoCs). In this way, various processing units are configured using one or more of the above-mentioned various processors as a hardware structure.

[0070] Furthermore, more specifically, the hardware structure of these various processors can be an electric circuit that combines circuit elements such as semiconductor elements.

[0071] The following are appendices to the present disclosure. (Additional note 1) at least one processor; The processor: acquiring medical information including date information associated with a medical procedure for the target patient and predicted date information for predicting a fall of the target patient; deriving fall prediction information for the target patient using a prediction model trained by machine learning to predict a fall for the target patient based on the date information and the number of days elapsed derived from the predicted date information; A medical support device that notifies the fall prediction information. (Additional note 2) 2. The medical support device according to claim 1, wherein the predicted date information is a planned hospital visit date for the target patient. (Additional note 3) The processor obtains the predicted date information; The medical support device according to claim 1, wherein the target patient is a patient associated with the predicted date information as a planned hospital visit date. (Additional note 4) The medical support device described in Appendix 1 or 2, wherein the date information is at least one of the date on which medication was prescribed for the target patient, the date on which an examination was performed on the target patient, the date on which surgery was performed on the target patient, the date on which the target patient was admitted to the hospital, the date on which the target patient was discharged from the hospital, and the date on which a diagnosis was made to the target patient. (Additional note 5) In the medical information, when a plurality of date information is associated with the medical treatment for the target patient, The processor: acquire specific date information that is closest to the predicted date information from among the plurality of date information; 5. A medical support device according to any one of claims 1 to 4, which derives the fall prediction information of the target patient using the prediction model based on the number of days elapsed derived from the specific date information and the predicted date information. (Additional note 6) The medical support device according to appended claim 5, wherein the fall prediction information is weighted based on the number of consecutive days that the medical procedure has been performed since the specific date information. (Additional note 7) The medical support device according to appended claim 5, wherein the fall prediction information is derived weighted based on the number of days on which medical procedures are performed within a predetermined period from the specific date information or the predicted date information. (Additional note 8) The medical support device according to any one of appendix 1 to 4, wherein the processor compares the number of elapsed days with a predetermined first threshold value and derives the fall prediction information based on the number of elapsed days being equal to or less than the first threshold value. (Additional note 9) 2. The medical support device according to claim 1, wherein the first threshold value is set to a different value for each classification of the medical information associated with the date information. (Additional note 10) the medical information includes at least one of basic patient information, disease information, medication information, surgery information, examination information, and keyword information; the date information is included in at least one of the patient basic information, the disease information, the medication information, the examination information, and the surgery information; The medical support device according to any one of appendix 1 to 5, wherein the processor derives the fall prediction information using the prediction model based on the number of days elapsed and at least one of the patient basic information, the disease information, the medication information, the surgery information, the examination information, and the keyword information. (Additional note 11) The processor derives a contribution of the medical information and the number of days elapsed input to the prediction model to the fall prediction information; The medical support device according to claim 10, wherein at least one of the medical information and the number of days elapsed is notified together with the fall prediction information based on the degree of contribution. (Additional note 12) the fall prediction information is at least one of whether or not there is a risk of falling and the probability of the risk of falling; determining whether the fall prediction information indicates that there is a fall risk or whether the fall risk is equal to or greater than a second threshold value; 12. The medical support device according to any one of claims 1 to 11, wherein the fall prediction information is notified based on the result of the determination. (Additional note 13) Item 13. The medical support device according to item 12, wherein the processor notifies a medical professional associated with the target patient of the fall prediction information. (Additional note 14) acquiring medical information including date information associated with a medical procedure for the target patient and predicted date information for predicting a fall of the target patient; deriving fall prediction information for the target patient using a prediction model trained by machine learning to predict a fall for the target patient based on the date information and the number of days elapsed derived from the predicted date information; A medical support method for notifying the fall prediction information. (Additional note 15) acquiring medical information including date information associated with a medical procedure for a target patient and predicted date information for predicting a fall of the target patient; deriving fall prediction information for the target patient using a prediction model trained by machine learning to predict a fall for the target patient based on the date information and the number of days elapsed derived from the predicted date information; and a procedure for notifying the patient of the fall prediction information. [Explanation of symbols]

[0072] 1 Medical support equipment 2 Management Server 3. Client terminal 4 Network 11 CPU 12 Medical Assistance Program 13 Storage section 14 Display 15 Input Devices 16 memory 17 Network I / F 18 RAM 19 Bus 21 Information Acquisition Department 22 Prediction Department 23 Notification Department 24 Predictive Models 30 Training data 31 days passed 32 Medical Information 33 Presence or absence of falls 40 List of high-risk patients 41 High-risk patient names 42 Patient Number 43 Reservation date and time 44 Contribution button 47 Contribution 50 Reservation Patient List 51 Patient name 52 Patient Number 53 Reservation date and time 54 Fall risk 55 marks

Claims

1. at least one processor; The processor: acquiring medical information including date information associated with a medical procedure for the target patient and predicted date information for predicting a fall of the target patient; deriving fall prediction information for the target patient using a prediction model trained by machine learning to predict a fall for the target patient based on the date information and the number of days elapsed derived from the predicted date information; A medical support device that notifies the fall prediction information.

2. The medical support device according to claim 1 , wherein the predicted date information is a planned hospital visit date of the target patient.

3. The processor obtains the predicted date information; The medical support device according to claim 1 , wherein the target patient is a patient associated with the predicted date information as a planned hospital visit date.

4. 3. The medical support device according to claim 1, wherein the date information is at least one of the date on which medication was prescribed for the target patient, the date on which an examination was performed for the target patient, the date on which surgery was performed for the target patient, the date on which the target patient was admitted to the hospital, the date on which the target patient was discharged from the hospital, and the date on which a diagnosis was made for the target patient.

5. In the medical information, when a plurality of date information is associated with the medical treatment for the target patient, The processor: acquire specific date information that is closest to the predicted date information from among the plurality of date information; The medical support device according to claim 1 or 2, wherein the fall prediction information for the target patient is derived using the prediction model based on the number of elapsed days derived from the specific date information and the predicted date information.

6. The medical support device according to claim 5 , wherein the fall prediction information is weighted based on the number of consecutive days that the medical procedure has been performed since the specific date information.

7. The medical support device according to claim 5 , wherein the fall prediction information is derived by weighting the fall prediction information based on the number of days on which medical procedures are performed within a predetermined period from the specific date information or the predicted date information.

8. The medical support device according to claim 1 or 2, wherein the processor compares the number of elapsed days with a predetermined first threshold value, and derives the fall prediction information based on the number of elapsed days being equal to or less than the first threshold value.

9. The medical support device according to claim 8 , wherein the first threshold value is set to a different value for each classification of the medical information associated with the date information.

10. the medical information includes at least one of basic patient information, disease information, medication information, surgery information, examination information, and keyword information; the date information is included in at least one of the patient basic information, the disease information, the medication information, the examination information, and the surgery information; 3. The medical support device according to claim 1, wherein the processor derives the fall prediction information using the prediction model based on the number of days elapsed and at least one of the patient basic information, the disease information, the medication information, the surgery information, the examination information, and the keyword information.

11. The processor derives a contribution of the medical information and the number of days elapsed input to the prediction model to the fall prediction information; The medical support device according to claim 10 , wherein at least one of the medical information and the number of elapsed days is notified together with the fall prediction information based on the degree of contribution.

12. the fall prediction information is at least one of whether or not there is a risk of falling and the probability of the risk of falling; determining whether the fall prediction information indicates that there is a fall risk or whether the fall risk is equal to or greater than a second threshold value; The medical support device according to claim 1 or 2, wherein the fall prediction information is notified based on the result of the determination.

13. The medical support device according to claim 12 , wherein the processor notifies a medical professional associated with the target patient of the fall prediction information.

14. acquiring medical information including date information associated with a medical procedure for the target patient and predicted date information for predicting a fall of the target patient; deriving fall prediction information for the target patient using a prediction model trained by machine learning to predict a fall for the target patient based on the date information and the number of days elapsed derived from the predicted date information; A medical support method for notifying the fall prediction information.

15. acquiring medical information including date information associated with a medical procedure for a target patient and predicted date information for predicting a fall of the target patient; deriving fall prediction information for the target patient using a prediction model trained by machine learning to predict a fall for the target patient based on the date information and the number of days elapsed derived from the predicted date information; and a procedure for notifying the patient of the fall prediction information.

Images