Patent Application
20250174365
The present disclosure relates to a non-transient computer readable medium, an information processing device, an information processing method, and a learning model generation method.
Japanese Patent Publication No. 2010-3135 A (“Patent Literature 1”) describes a medical information management system that includes a plurality of medical department servers that provides medical information stored in a database of a plurality of medical departments in a medical institution and a specific disease treatment integrated server that acquires specific disease medical information from the medical department server and provides the integrated specific disease medical information from an integrated database, in which a treatment progressing status can be grasped with high immediacy by making it possible to refer to medical information of specific diseases such as cancer in all treatment methods combining surgery, radiation therapy, chemotherapy, and palliative care and registered related information of the specific diseases in a wide range.
For diseases such as cancer, there is a method of short-term prognosis prediction (for example, palliative prognostic index (PPI)). However, this is prognosis regarding a medical condition and is not focused on pain caused by the disease. Furthermore, there is a problem that a doctor can evaluate pain only when the doctor faces a patient. Furthermore, specialized knowledge is essential for effective pain management, and there is a problem that it is difficult for a doctor other than a specialist to perform an accurate pain treatment according to the patient.
Embodiments of present invention have been developed in view of the above circumstances, and an object of certain embodiments of the present invention is to provide a non-transient computer-readable medium, an information processing device, an information processing method, and a learning model generation method that can assist an appropriate pain treatment.
According to one embodiment, a non-transient computer readable medium comprises a computer program that causes a computer to execute processing comprising: acquiring pain related information regarding pain of a patient; inputting the acquired pain related information into a learning model that outputs pain prediction information, and acquiring, from the learning model, pain prediction information of the patient; generating pain treatment assistance information for the patient based on the acquired pain prediction information; and outputting the generated pain treatment assistance information.
According to one aspect, the pain related information includes at least one of an evaluation index used to evaluate pain, vital data, medication information, and a physical condition.
According to one aspect, the computer program causes the computer to execute processing further comprising: acquiring medical information regarding medical care of the patient; and inputting the acquired medical information into the learning model that outputs the pain prediction information, and acquiring, from the learning model, the pain prediction information of the patient.
According to one aspect, the medical information includes at least one of diagnosis information, examination information, treatment information, prescription information, and disease prognosis prediction information.
According to one aspect, the computer program causes the computer to execute processing further comprising: acquiring literature information regarding a pain treatment; and inputting the acquired literature information into the learning model that outputs the pain prediction information, and acquiring, from the learning model, the pain prediction information of the patient.
According to one aspect, the pain treatment assistance information includes at least one of recommended prescription information and recommended treatment information used to assist a doctor who examines the patient.
According to one aspect, the pain treatment assistance information includes a coping method for assisting the patient.
According to one aspect, the computer program causes the computer to execute processing further comprising: outputting a pain prediction pattern indicating a temporal transition of a degree of pain based on the pain prediction information of the patient.
According to another embodiment, a control unit is programmed to execute steps comprising: acquiring pain related information regarding pain of a patient; inputting the acquired pain related information into a learning model that outputs pain prediction information, and acquiring, from the learning model, pain prediction information of the patient; generating pain treatment assistance information for the patient based on the acquired pain prediction information; and outputting the generated pain treatment assistance information.
According to another embodiment, an information processing method comprises: acquiring pain related information regarding pain of a patient; inputting the acquired pain related information into a learning model that outputs pain prediction information, and acquiring, from the learning model, pain prediction information of the patient; generating pain treatment assistance information for the patient based on the acquired pain prediction information; and outputting the generated pain treatment assistance information.
According to another embodiment, a learning model generation method comprises: acquiring first training data including pain related information regarding pain of a plurality of patients and pain prediction information of the plurality of patients; and generating a learning model that receives pain related information of a patient and outputs pain prediction information of the patient based on the acquired first training data.
According to one aspect, the method further comprises: acquiring second training data further including medical information of the plurality of patients, literature information regarding a pain treatment, and the pain prediction information of the plurality of patients; and generating the learning model so as to output the pain prediction information based on the acquired second training data.
According to certain embodiments of the present invention, an appropriate pain treatment can be assisted.
Hereinafter, an embodiment of the present invention will be described.
The patient terminal device 10 is a terminal device carried or held by a patient, and the assistant terminal device 30 is a terminal device carried or held by an assistant such as a family member or a caregiver of the patient. The caregiver may be a family member or a caregiver at a care provider. The patient terminal device 10 and the assistant terminal device 30 can include a smartphone, a tablet, a personal computer, or the like including a display panel, an operation panel, a microphone, a speaker, or the like. In the patient terminal device 10 and the assistant terminal device 30, an app (application) for using the pain treatment assistance system is introduced. For example, a patient who is suffering from cancer or the like and periodically visits a medical institution while being treated at home is assumed. However, the patient is not limited to such a patient. Note that the app may be a Web app or a downloaded app.
The device 20 is a measuring instrument that measures vital data of the patient and may be a wearable terminal or a stationary type. The device 20 can measure the vital data such as a body temperature, a heart rate, brain waves, or electrodermal activity (EDA). Note that the vital data is not limited to these, and may include any data assumed to be related to pain caused by the patient's disease. For example, a blood pressure, a pulse pressure, pulse waves, a blood sugar level, a weight, the number of steps, an activity amount, or the like may be included in the vital data.
The doctor terminal device 40 is a terminal device used by a doctor in a medical institution or the like. The doctor terminal device 40 can include a personal computer or the like including a display panel, an operation panel, or the like. A medical worker such as a public health nurse or a nurse may use the doctor terminal device 40 under guidance of the doctor. An application (for example, Web app or the like) for using the pain treatment assistance system is introduced in the doctor terminal device 40.
The storage unit 59 can include, for example, a hard disk, a semiconductor memory, or the like, and stores a computer program 60 (program product), a learning model 61, and necessary information. The computer program 60 may be downloaded from an external device via the communication unit 52 and stored in the storage unit 59. Furthermore, the computer program 60 recorded in a recording medium (for example, optically readable disk storage medium such as CD-ROM) may be read by a recording medium reading unit and stored in the storage unit 59, or the computer program 60 may be read by the recording medium reading unit and developed in the memory 53.
The control unit 51 may be configured by incorporating a required number of central processing units (CPUs), micro-processing units (MPUs), graphics processing units (GPUs), or the like. The control unit 51 can execute processing defined by the computer program 60. That is, the processing executed by the control unit 51 corresponds to processing executed in accordance with the computer program 60. The control unit 51 can execute functions of the assistance information generation unit 58, by executing the computer program 60. The assistance information generation unit 58 may include hardware or software or may be implemented by a combination of hardware and software. The control unit 51 can execute processing using the learning model 61. Note that the control unit 51 has functions as a first acquisition unit, a second acquisition unit, a generation unit, and an output unit.
The communication unit 52 includes a communication module and can exchange necessary information with the patient terminal device 10, the assistant terminal device 30, and the doctor terminal device 40 via the communication network 1. Furthermore, the communication unit 52 may exchange necessary information with the device 20.
The memory 53 may include a semiconductor memory, such as a static random access memory (SPAM), a dynamic random access memory (DRAM), or a flash memory. Loading the computer program 60 into the memory 53 enables the control unit 51 to execute the computer program 60.
Furthermore, the pain treatment assistance device 50 acquires medical information related to medical care of the patient from the doctor terminal device 40 via the communication unit 52. The acquired medical information is classified for each patient and for each medical care time point and stored in the medical information DB 55.
The pain evaluation index includes an intensity of pain, a change in the pain, a timing of the pain, or the like. The pain evaluation index can be transmitted from the patient terminal device 10 to the pain treatment assistance device 50, for example, when the patient activates the app of the patient terminal device 10 and answers a question regarding the intensity of the pain, the change in the pain, the timing of the pain, or the like.
The vital data includes the body temperature, the heart rate, the brain waves, the EDA, the pulse waves, or the like that changes in response to the pain. Note that the blood pressure, the pulse pressure, the blood sugar level, the weight, the number of steps, the activity amount, or the like may be included. The vital data can be acquired using the device 20.
The medication information includes a medicine type, a dosage, a medicine taking period, a medicine taking time (timing), whether or not to take the medicine, or the like. The medication information may include not only a medicine that is currently taken but also a medicine that has been taken in the past.
The physical condition includes an activity amount, a sleep time, sleepiness, lassitude, delirium, constipation, voice, facial expression, or the like. It is possible to include an item that can evaluate whether or not a disorder occurs due to the pain, the medicine, medical conditions, or the like.
The pain related information includes at least one of the pain evaluation index (evaluation index used to evaluate pain), the vital data, the medication information, and the physical condition.
By acquiring the pain related information as described above from the patient terminal device 10 and the device 20, the patient can accurately tell pain and symptoms that the patient feels at a necessary timing and a necessary frequency, in a period from a previous examination to a next examination.
The diagnosis information includes findings by interview, inspection, palpation, or percussion at the time of examination, an electronic chart, or the like. The examination information includes a blood test, a urine test, an image test, or the like. The prescription information includes a medicine type and a dosage prescribed by a doctor. Furthermore, the medicine taking period may be included. The treatment information includes a treatment method and a treatment period. Note that not only a literal treatment but also medicine prescription are included in a pain treatment. The disease prognosis prediction information includes a palliative prognostic index (PPI). The PPI is a representative index as an index used to predict a short-term life prognosis (in weeks), and a score obtained by adding scores of a palliative performance scale (PPS), an ingestion dose, an edema, resting dyspnea, and a delirium. As the score increases, predicted prognosis is shorter.
The medical information includes at least one of the diagnosis information, the examination information, the treatment information, the prescription information, and the disease prognosis prediction information.
The learning model 61 is generated (learned) to output pain prediction information (pain prediction information) in a case where the pain related information is input. The pain related information is time-series data and includes, for example, pain related information within a period of about one month, three months, six months, or 12 months in the past going back from the present. The pain prediction information is information indicating pain within a required period from the current point to the future (for example, one month, three months, six months, 12 months, or the like). The learning model 61 may, for example, output the pain prediction information from the present to one month later based on the pain related information for past three months or may output the pain prediction information from the present to three months later, based on the pain related information for past six months. Note that these periods are examples, and the present invention is not limited to these. The degree (degree) of the pain can be expressed, for example, in several stages such as 10 stages from one to 10 or five stages from one to five. It can be assumed that the pain of the patient be stronger as a numerical value increases.
The learning model 61 can use, for example, a support vector machine (SVMV), a decision tree, a random forest, an adaboost, a neural network (for example, recurrent neural network (RNN), long short term memory (LSTM), a transformer, or the like), or the like.
The control unit 51 reads a past pain pattern of the patient, recorded in the pain treatment assistance information DB 56. The pain pattern is information regarding pain in the past, and is information indicating a transition of a degree of pain, similarly to the pain prediction information. A difference between the pain prediction information and the past pain pattern is that the former is a transition of a degree of pain in a period from the present to the future, whereas the latter is a transition of a degree of pain in a period from the present to the past. The control unit 51 inputs the read past pain pattern and the pain prediction information acquired from the learning model 61, into the assistance information generation unit 58.
The assistance information generation unit 58 generates the pain treatment assistance information based on the pain prediction information and the past pain pattern that have been input. The assistance information generation unit 58 stores the generated pain treatment assistance information in the pain treatment assistance information DB 56. The pain treatment assistance information includes, for example, a pain prediction pattern, recommended prescription information and recommended treatment information used to assist the doctor who examines the patient, recommended coping information used to assist the patient, or the like. It is sufficient that the pain treatment assistance information include at least one of the recommended prescription information and the recommended treatment information used to assist the doctor.
It is sufficient that the pain prediction pattern be, for example, information that is generated based on the pain prediction information and indicates a temporal transition of a degree of pain of the patient from the past to the future. The pain prediction pattern can be displayed by an appropriate method such as graph display or tabular display. The recommended prescription information is information for assisting the doctor and includes information used to determine recommended prescription in medicine prescription for the patient. The recommended treatment information is information for assisting the doctor and includes information used to determine a recommended treatment method among treatment methods other than the medicine prescription for the patient. The recommended coping information is information for assisting the patient and, for example, is information that enables the patient side to cope with the pain even in a case where there is no doctor's intervention such as the examination by the doctor, by notifying at least one of the patient or the family member or the caregiver in advance of the information in a case where the pain is expected in the future.
The assistance information generation unit 58 may generate the pain treatment assistance information on a rule basis, for example, using a table indicating a correspondence relationship between the pain prediction information and the past pain pattern, and the pain treatment assistance information. The table may indicate the correspondence relationship between the pain prediction information and the past pain pattern, and the pain treatment assistance information.
Furthermore, the assistance information generation unit 58 can be configured by a model using a machine-learned algorithm and can use, for example, a support vector machine (SVMV), a decision tree, a random forest, an adaboost, a neural network, or the like.
As described above, the control unit 51 can acquire the pain related information regarding the pain of the patient, input the acquired pain related information into the learning model 61 that outputs the pain prediction information in a case where the pain related information is input and acquire the pain prediction information of the patient, generate the pain treatment assistance information for the patient based on the acquired pain prediction information, and output the generated pain treatment assistance information.
Furthermore, the control unit 51 reads the literature information from the literature information DB 57 and inputs the read literature information into the learning model 61. The control unit 51 can perform the vectorization operation for generating the primary vector for the literature information, generate a literature information vector, and input the generated literature information vector into the learning model 61. It is sufficient that the literature information include, for example, information necessary for associating the type of history information such as the symptom history of the patient, the history of the vital data, the examination history, the medication history, or the treatment history, with how the situation of the pain of the patient changes in the future. For example, if there is a paper indicating how pain changes in the future when a specific medicine is taken, the paper can be used as the literature information. Furthermore, if there is a guideline indicating how the pain changes in the future when a specific treatment is performed, the guideline can be used as the literature information. In particular, by compensating shortage of the pain related information and the medical information with the literature information, in a case where the pain related information of the patient and the medical information are insufficient (for example, in a case where past data is small), the learning model 61 can output the pain prediction information of the patient. It is expected that accuracy of an index output by the learning model 61 is improved by increasing the number of types of information to be input into the learning model 61.
Furthermore, the control unit 51 reads the literature information from the literature information DB 57 and inputs the read literature information into the learning model 61. The control unit 51 can perform the vectorization operation for generating the primary vector for the literature information, generate a literature information vector, and input the generated literature information vector into the learning model 61. It is expected that accuracy of an index output by the learning model 61 is improved by increasing the number of types of information to be input into the learning model 61.
As described above, in a case of acquiring the medical information regarding the medical care of the patient and further inputting the medical information, the control unit 51 may input the acquired medical information into the learning model 61 that outputs the pain prediction information and acquire the pain prediction information of the patient.
Furthermore, in a case of acquiring the literature information regarding the pain treatment and further inputting the literature information regarding the pain treatment, the control unit 51 may input the acquired literature information into the learning model 61 that outputs the pain prediction information and acquire the pain prediction information of the patient.
The assistance information generation unit 58 may generate the pain treatment assistance information on a rule basis, for example, using a table indicating a correspondence relationship between the pain prediction information, the past pain pattern, the pain related information of the patient, the medical information, and the literature information, and the pain treatment assistance information. The table may indicate the correspondence relationship between the pain prediction information, the past pain pattern, the pain related information of the patient, the medical information, and the literature information, and the pain treatment assistance information.
Furthermore, as in the first and second examples, the assistance information generation unit 58 can be configured by the model using the machine-learned algorithm and can use, for example, a support vector machine (SVMV), a decision tree, a random forest, an adaboost, a neural network, or the like.
Note that, in the example in
Next, a specific example of the pain treatment assistance information will be described. The control unit 51 can output the pain treatment assistance information to the doctor terminal device 40, the patient terminal device 10, or the assistant terminal device 30. The pain treatment assistance information is displayed on the display panel of the doctor terminal device 40, and the doctor can perform an input operation or a selection operation according to displayed content. Similarly, the pain treatment assistance information is displayed on the display panel of the patient terminal device 10 or the assistant terminal device 30, and the patient, the family member or the caregiver, or the like can perform the input operation or the selection operation according to the displayed content.
Furthermore, as an advice (assistance information) to the doctor, “Pain may occur in future” is displayed. When the doctor operates a “recommended prescription information” icon 401, it is possible to display the recommended prescription information to be described later.
As described above, the control unit 51 outputs the pain prediction pattern of the patient to the doctor terminal device 40.
As described above, a pain occurrence frequency and period in the future can be predicted according to the pain prediction pattern of the patient, and the doctor can present analgesics suitable for the patient, in advance. When the patient takes the predicted analgesics, the pain predicted to occur in the future does not occur, and it is possible to improve quality of life (QOL) of the patient. In this way, it is possible to assist an appropriate pain treatment.
Furthermore, as an advice (assistance information) to the doctor, “Pain seems to continue in future. How about considering new treatment method” is displayed. When the doctor operates a “recommended treatment information” icon 421, the recommended treatment information to be described later can be displayed.
Furthermore, as an advice (assistance information) to the patient or an assistant, “Occurrence of pain is predicted. Please check coping method” is displayed. When the patient or the assistant operates a “check” icon 101, a coping method to be described can be displayed.
As described above, the control unit 51 outputs the pain prediction pattern of the patient to the patient terminal device 10 and the assistant terminal device 30.
As described above, according to the pain prediction pattern of the patient, the patient can know an occurrence frequency and period of the pain in the future, in advance. Then, by performing necessary coping methods, according to the coping method recommended by the pain treatment assistance device 50, it is possible to prevent the occurrence of the pain in advance, and the quality of life (QOL) of the patient can be improved. In particular, for example, the patient can perform the necessary coping method at home, without visiting the hospital. In this way, it is possible to assist an appropriate pain treatment. Note that, in a case where it is not preferable to directly notify the patient of the pain treatment assistance information on the patient side, it is sufficient to transmit the pain treatment assistance information to the assistant terminal device 30, instead of the patient terminal device 10.
As described above, according to the present embodiment, it is possible to predict pain that may occur in the patient in the future, from the state of the patient, and it is possible to perform the pain treatment for alleviating the pain on the patient in advance. Since preventive measures can be taken in this way, the pain of the patient can be reduced.
The control unit 51 acquires the medical information of the patient (S13) and stores the acquired medical information in the medical information DB 55 (database) (S14). The medical information can be acquired not only at the time of examination but also when the doctor determines that the medical information is needed.
The control unit 51 inputs the pain related information of the patient, the medical information, and the literature information regarding the pain treatment into the learning model 61 and acquires the pain prediction information of the patient, output by the learning model 61 (S15). The control unit 51 reads the past pain pattern of the patient from the pain treatment assistance information DB 56 (database) (S16).
The control unit 51 generates the pain treatment assistance information including the pain prediction pattern of the patient, based on the past pain pattern of the patient read from the pain treatment assistance information DB 56 and the pain prediction information of the patient acquired from the learning model 61 (S17). In this case, the pain treatment assistance information is generated based on the pain related information of the patient, the medical information, and the literature information, in addition to the past pain pattern and the pain prediction information.
The control unit 51 stores the generated pain treatment assistance information in the pain treatment assistance information DB 56 (database) (S18), outputs the generated pain treatment assistance information (S19), and ends the processing.
| Number | Date | Country | Kind |
|---|---|---|---|
| 2022-118143 | Jul 2022 | JP | national |
This application is a bypass continuation of PCT Application No. PCT/JP2023/022147, filed on Jun. 14, 2023, which claims priority to Japanese Patent Application No. 2022-118143, filed on Jul. 25, 2022. The entire contents of these application are incorporated herein by reference.
| Number | Date | Country | |
|---|---|---|---|
| Parent | PCT/JP2023/022147 | Jun 2023 | WO |
| Child | 19035631 | US |
