JP7313170B2 - Medical information processing apparatus, medical information processing system, and medical information processing method - Google Patents

Description

TECHNICAL FIELD Embodiments of the present invention relate to a medical information processing apparatus, a medical information processing system, and a medical information processing method.

With advances in computer processing power, simulation-based estimation techniques are being applied to the medical and healthcare fields. Simulations include a variety of simulations, from physical simulations such as computational fluid dynamics analysis and fluid-structure interaction to bio-simulations. Estimation technology based on simulation models will further advance in the future, and is considered an important technology that can contribute to the realization of future personalized medicine, such as estimating disease progress and treatment outcomes.

On the other hand, along with the spread of IoT technology, knowledge-based estimation technology based on big data (estimation technology based on trained models) has made great progress. For this reason, it has become possible to collect a large amount of extremely diverse types of medical data from subjects. For example, this type of medical data includes diagnostic imaging results and examination information obtained in hospitals, biological information obtained outside hospitals and at home, and information from wearable sensors.

However, these two estimation techniques have evolved independently, and have increased their application cases independently in their respective fields of expertise. Therefore, it is difficult to use them in combination.

In addition, estimation using a trained model and estimation using a simulation model have different problems, and it is difficult to uniformly determine which is superior. For example, it may be difficult to set boundary conditions in simulation model estimation. In this case, estimation using a simulation model is performed on an ad-hoc basis by roughly setting probable numerical values as boundary conditions and performing comparisons and matching with experiments and standard models, which may result in large errors in estimation results. In addition, the accuracy of the estimation result by the trained model is affected by the type, quality, and quantity of the input data, but the measurement information obtained from the individual is limited, and it is difficult to obtain all the information to reproduce the biological information of the subject. For this reason, it is difficult to perform estimation that reflects individual characteristics and feature amounts in both estimation using a trained model and estimation using a simulation model.

Japanese Patent Publication No. 2014-532859 Christopher M. Bishop, "Pattern recognition and machine learning", (USA), 1st ed., Springer, 2006, p. 225-290

The problem to be solved by the present invention is to estimate the health condition of a subject by combining estimation by a trained model and estimation by a simulation model.

A medical information processing apparatus according to an embodiment includes an acquisition unit and an optimization unit. The acquisition unit acquires a correct data set from a storage unit that stores a correct data set including sets of verification medical data collected from a plurality of subjects and correct estimation results of information on health conditions. The optimization unit determines which of the trained model and the simulation model is used to estimate the information about the health condition of the subject to be estimated, based on the results of estimating the information about the health condition generated by the learned model based on the medical data for verification, the results of estimating the information about the health condition generated by the simulation model based on the medical data for verification, and the results of estimating correct information about the health condition.

1 is a block diagram showing a configuration example of a medical information processing system including a medical information processing apparatus according to an embodiment; FIG. FIG. 2 is a block diagram showing a configuration example of an information processing terminal; 4 is a flowchart showing an example of an optimization process procedure for determining which of a trained model and a simulation model is used to estimate information about the health condition of a subject to be estimated, which is performed for each estimation process. 4 is a flowchart showing an example of a procedure for optimizing each estimation process when inputs and outputs of three estimation processes can be connected in series. 4 is a flow chart showing an example of a schematic procedure for constructing a trained model and a simulation model for each estimation process; 4 is a flowchart showing an example of a procedure when estimation processing is operated by a simulation model and part of the input medical data necessary for estimation processing is insufficient. 4 is a flowchart showing an example of a specific procedure for constructing a trained model and a simulation model for estimating coronary artery blood flow parameters. 4 is a flow chart showing an example of a procedure when a process for estimating a blood flow parameter of a coronary artery is operated by a simulation model X and part of the input medical data necessary for the estimation process is insufficient. FIG. 10 is a flowchart showing an example of a procedure when estimation processing is operated using a trained model and part of the input medical data necessary for estimation processing is insufficient; FIG.

Hereinafter, embodiments of a medical information processing apparatus, a medical information processing system, and a medical information processing method will be described in detail with reference to the drawings.

(Overall system configuration)
FIG. 1 is a block diagram showing a configuration example of a medical information processing system 10 including a medical information processing apparatus 20 according to one embodiment. FIG. 2 is a block diagram showing a configuration example of the information processing terminal 40. As shown in FIG.

The medical information processing system 10 includes a medical information processing apparatus 20 as an example of a server, a correct answer database (correct answer DB) 31, a subject database (subject DB) 32, a plurality of modalities 35 and 36, an information processing terminal 40 used by a user such as a doctor, a wearable terminal 51 attached to the subject, an IoT gateway 52, and an information processing terminal 53 used by the subject.

The correct answer DB 31 and the subject DB 32 have a configuration including a processor-readable recording medium such as a magnetic or optical recording medium or a semiconductor memory.

The correct answer DB 31 stores a correct answer data set consisting of a set of verification medical data collected from a plurality of subjects and an estimation result of the correct answer of the information on the health condition. Verification medical data is data collected in the past from a plurality of subjects. The estimated result of the correctness of the information on the health condition is, for example, a definitive diagnosis result regarding a lesion determined by diagnosing a tissue extracted from the person to whom the verification medical data corresponding to the estimation is collected. The correct answer DB 31 is an example of a storage unit.

The subject DB 32 stores medical data collected from subjects for each subject. A PHR (Personal Health Record) or the like can be used as the subject DB 32 . The medical data collected in the subject DB 32 include various medical data (including medical image data collected by the modalities 35 and 36) collected when the subject visits the hospital, medical data collected by the wearable terminal 51 attached to the subject, or medical data collected by a device (not shown) possessed by the subject (medical measuring equipment, such as a sphygmomanometer) that can be connected to a network. The subject DB 32 is an example of a storage unit.

The plurality of modalities 35 and 36 are composed of, for example, nuclear medicine diagnostic apparatuses such as an ultrasound diagnostic apparatus, an X-ray CT (Computed Tomography) apparatus, an MRI (Magnetic Resonance Imaging) apparatus, a SPECT (Single Photon Emission computed Tomography) apparatus, and a PET (Positron Emission Tomography) apparatus.

The information processing terminal 40 is used by a user such as a doctor, and is composed of, for example, a general personal computer or workstation. The information processing terminal 40 may be configured by a portable information processing terminal such as a smart phone type, tablet terminal, notebook type personal computer, or the like.

As shown in FIG. 2, the information processing terminal 40 has a display 41 , an input interface 42 , a memory circuit 43 , a communication circuit 44 and a processing circuit 45 . The information processing terminal 40 is an example of a client that displays the estimation results output by the medical information processing apparatus 20 .

The display 41 is composed of a general display output device such as a liquid crystal display or an OLED (Organic Light Emitting Diode) display. The input interface 42 is composed of general input devices such as a trackball, switch button, mouse, keyboard, numeric keypad, etc., and outputs operation input signals corresponding to user's operations to the processing circuit 45 .

The storage circuit 43 includes a processor-readable recording medium such as a RAM (Random Access Memory), a semiconductor memory device such as a flash memory, a hard disk, an optical disk, or the like, and stores programs, parameter data, and other data used by the processing circuit 45. Part or all of the programs and data in the recording medium of the storage circuit 43 may be downloaded by communication via the network 100, or may be provided to the storage circuit 43 via a portable storage medium such as an optical disk.

Communication circuit 44 implements various information communication protocols according to the form of network 100 . The communication circuit 44 connects with other electric devices via the network 100 according to these various protocols. Communication circuit 44 is an example of a receiver.

In the present embodiment, the network 100 means, for example, a cloud network or any other general information communication network using telecommunication technology, and includes a wireless/wired LAN such as a hospital backbone LAN (Local Area Network), an Internet network, a telephone communication network, an optical fiber communication network, a cable communication network, a satellite communication network, and the like.

The processing circuit 45 implements a function of centrally controlling the information processing terminal 40 . The processing circuit 45 is a processor that reads out and executes a program stored in the storage circuit 43 to execute a process of requesting the medical information processing apparatus 20 to estimate information on the health condition, and a process of receiving the estimated result of the information on the health condition from the medical information processing apparatus 20 and displaying it on the display 41.

The processor of the processing circuit 45 implements an estimation request function 451 , an estimation result acquisition function 452 and an image generation function 453 . Each of these functions is stored in the storage circuit 43 in the form of a program.

The estimation request function 451 requests the medical information processing apparatus 20 to estimate information regarding the health condition of the subject.

The estimation result acquisition function 452 acquires the estimation result of information on the health condition of the subject from the medical information processing apparatus 20 received by the communication circuit 44 .

The image generation function 453 generates an image for supporting diagnosis of the subject based on the estimation result of the information regarding the health condition of the subject acquired from the medical information processing apparatus 20 and causes the display 41 to display the image. The images for supporting the diagnosis of the subject include images showing the result of estimating the information about the health condition of the subject itself.

The wearable terminal 51 (see FIG. 1) is a medical data collection device that is worn by a subject on a part of the body such as an arm, and collects medical data of the subject. The medical data collected by the wearable terminal 51 includes various biochemical information such as blood pressure, heart rate, blood oxygen concentration, blood sugar level, and intensity of ultrasonic waves, infrared rays, and the like. When the wearable terminal 51 has a display, the wearable terminal 51 is an example of a client that displays the estimation results output by the medical information processing apparatus 20 .

The IoT gateway 52 has a processor, performs edge processing on medical data transmitted from the wearable terminal 51 , and then transmits the medical data to the network 100 . Edge processing is processing performed to reduce the amount of data transmitted from wearable terminal 51, and includes, for example, processing for reducing the amount of data transmitted to network 100 based on the feature amount extracted from the data.

Although FIG. 1 shows an example in which the wearable terminal 51 is connected to the network 100 via the IoT gateway 52 , it may be connected directly to the network 100 without the IoT gateway 52 . When the wearable terminal 51 is connected to the network 100 via the IoT gateway 52, even if collected data is always transmitted in real time from the wearable terminal 51, the edge processing of the IoT gateway 52 can reduce the amount of data flowing to the network 100.

The information processing terminal 53 is used by the subject, and is configured by a portable information processing terminal such as a smartphone type, tablet terminal, or notebook type personal computer. The information processing terminal 53 is an example of a client that displays the estimation results output by the medical information processing apparatus 20 .

(Configuration of medical information processing apparatus)
The medical information processing apparatus 20 is composed of a general information processing apparatus such as a personal computer or workstation, and has an input interface 21, a display 22, a storage circuit 23, a communication circuit 24, and a processing circuit 25 as shown in FIG.

The input interface 21 of the medical information processing apparatus 20 is composed of general input devices such as a trackball, switch buttons, mouse, keyboard, numeric keypad, etc., and outputs operation input signals corresponding to user's operations to the processing circuit 25 . The display 22 is composed of a general display output device such as a liquid crystal display or an OLED display.

The storage circuit 23 has a configuration including a processor-readable recording medium such as a semiconductor memory device such as a RAM, a flash memory, a hard disk, an optical disk, etc., and stores programs, parameter data, and other data used by the processing circuit 25. Part or all of the programs and data in the recording medium of the storage circuit 23 may be downloaded by communication via the network 100, or may be provided to the storage circuit 23 via a portable storage medium such as an optical disc.

Communication circuit 24 implements various information communication protocols according to the form of network 100 . The communication circuit 24 connects with other electric devices via the network 100 according to these various protocols. Communication circuit 24 is an example of a transmitter.

The processing circuit 25 implements a function of centrally controlling the medical information processing apparatus 20 . The processing circuit 25 is a processor that reads out and executes the medical information processing program stored in the storage circuit 23 to perform processing for estimating the health condition of the subject by combining estimation by the trained model and estimation by the simulation model.

The processor of processing circuit 25 implements acquisition function 251 , estimation function 252 , optimization function 253 , learning function 254 and notification function 255 . Each of these functions is stored in the storage circuit 23 in the form of a program.

The acquisition function 251 acquires, from the correct answer DB 31 , a correct answer data set consisting of a set of verification medical data collected from a plurality of subjects and an estimated correct answer result of information on health conditions. Also, the acquisition function 251 acquires medical data of the subject to be estimated from the subject DB 32 . Acquisition function 251 is an example of an acquisition unit.

The estimation function 252 uses either the learned model determined by the optimization unit or the simulation model to estimate information regarding the health condition of the subject to be estimated based on the medical data of the subject to be estimated. The estimation function 252 is an example of an estimation unit.

The estimation process according to this embodiment roughly includes three processes. One is a process of estimating current medical parameters of a subject using medical data of the subject as input data. Medical parameters include, for example, blood flow parameters of coronary arteries. One is a process of estimating the current health condition of a subject using the medical data of the subject as input data. The current health condition includes the name of a disease that the subject is currently suffering from. One is a process of estimating a change in the subject's future health condition, an estimation of prognosis, or a process of estimating a treatment plan, using the medical data of the subject as input data. Health status information includes current medical parameters, current health status, and future health status estimates or treatment plans. Future health status estimation includes future health status change prediction and prognosis prediction.

The optimization function 253 determines which of the trained model and the simulation model is used to estimate the information about the health condition of the subject to be estimated based on the estimation result of the information about the health condition generated by the trained model based on the medical data for verification, the result of estimating the information about the health condition generated by the simulation model based on the medical data for verification, and the result of estimating the correctness of the information about the health condition. The optimization function 253 is an example of an optimization unit.

In addition, while it is important that the estimation result is close to the correct answer, there are situations where it is important that the estimation result can be obtained in a short time. In particular, estimation processing using a simulation model that requires complicated calculations may take an unexpectedly long time and may not be practical. Therefore, the optimization function 253 may determine which of the trained model and the simulation model to use, further based on the time required for each of the trained model and the simulation model to generate an estimation result.

This optimization process is performed for each estimation process, for example, for each of the above three estimation processes. Further, each of the three estimation processes described above may be subdivided into a plurality of estimation processes, and in this case, an optimization process may be performed as to which of the trained model and the simulation model is used for each of the subdivided estimation processes.

The learning function 254 pre-builds a trained model and a simulation model for each estimation process. Learning function 254 is an example of a learning unit.

The learning function 254 can build trained models using deep learning using multilayer neural networks, such as CNNs (Convolutional Neural Networks) and Convolutional Deep Belief Networks (CDBNs). Various methods, such as the method disclosed in Non-Patent Document 1, are known as methods of learning and methods of constructing a trained model.

Note that the trained model and the simulation model may each be constructed by an integrated circuit such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array).

The notification function 255 notifies at least one of the user and the subject of the estimation result of the information regarding the health condition of the subject estimated by the estimation function 252 . For example, the notification function 255 displays an image indicating information about the health condition of the subject generated by the estimation function 252 on the display 41 of the information processing terminal 40 used by the user, the display of the wearable terminal 51 used by the subject, the display of the information processing terminal 53, or the like. More specifically, the notification function 255 can display colors and shapes, text information, numerical values, and graphs indicating the current health condition of the subject on a web page on the Internet. In addition, the notification function 255 may cause these terminals to output audio indicating information about the health condition of the subject. A user such as a doctor can easily and accurately diagnose the health condition of the subject by confirming the information about the health condition of the subject notified by the notification function 255 as a message, numerical value, graph information, etc. based on the estimation result by the estimation function 252. For example, if the health status information is a result of an estimate of a patient's response to a therapy, then the therapeutic agent can be administered after the patient's response to the therapy has been predicted.

(optimization)
FIG. 3 is a flowchart showing an example of an optimization process procedure for determining which of the trained model and the simulation model is used to estimate information about the health condition of the subject to be estimated, which is performed for each estimation process. In FIGS. 3-9, numerals attached to S indicate respective steps of the flow chart. FIG. 3 shows an example of optimization processing for one estimation processing.

Note that the learned model and the simulation model are built in advance by the learning function 254 for each estimation process.

First, in step S<b>1 , the acquisition function 251 acquires a correct data set (a set of verification medical data and a correct estimation result of information on the health condition) from the correct DB 31 .

Next, in step S2, the estimating function 252 generates an estimated result of health condition information using a trained model based on the verification medical data.

Next, in step S3, the estimating function 252 generates an estimated result of health condition information using a trained model based on the verification medical data. Since one estimation process is assumed now, the input data in the estimation processes in step S2 and step S3 are the same, and the output information regarding the health condition is also of the same type.

Next, in step S4, the optimization function 253 determines which model is to be used for the estimation process based on the estimation result of each of the learned model and the simulation model and the correct estimation result.

By the above procedure, it is possible to determine for each estimation process whether to use the learned model or the simulation model to estimate the information about the health condition of the subject to be estimated.

The procedure shown in FIG. 3 can be applied, for example, to each of the above three estimation processes. Also, in the above description, each of the three estimation processes uses the medical data of the subject as input data, but the inputs and outputs of the three estimation processes may be connected in series.

FIG. 4 is a flowchart showing an example of a procedure for optimizing each estimation process when the inputs and outputs of three estimation processes can be connected in series.

In step S11, the optimization function 253 determines whether to use the learned model or the simulation model according to the procedure shown in FIG.

Next, in step S12, the optimization function 253 determines whether to use the learned model or the simulation model according to the procedure shown in FIG.

Next, in step S13, the optimization function 253 determines whether to use the learned model or the simulation model according to the procedure shown in FIG.

By the above procedure, each estimation process can be optimized when the inputs and outputs of the three estimation processes can be connected in series. According to the procedure shown in FIG. 4, when a plurality of estimation processes are related to each other and the final estimation result is output, for example, a combination of using a trained model for the first estimation process and the second estimation process and using a simulation model for the last estimation process can be used.

Also, the optimization function 253 may optimize a combination of models responsible for each of a plurality of estimation processes using trained models. This trained model is constructed to fit the estimation of the combination of models. In this case, the optimization function 253 inputs information on a plurality of estimation processes to a trained model for combined estimation that generates a combination of models for each estimation process based on information on a plurality of estimation processes, thereby determining a combination of models for each of the plurality of estimation processes.

(learning, operation)
First, an outline of model construction and model operation will be described with reference to FIGS. 5 and 6, respectively.

FIG. 5 is a flow chart showing an example of a schematic procedure for constructing a trained model and a simulation model for each estimation process.

In step S21, the learning function 254 constructs a simulation model for estimating information about health conditions based on medical data. A plurality of pieces of medical data may be input.

In step S22, the learning function 254 builds a trained model for estimating other medical data based on the medical data. A plurality of pieces of medical data may be input. The medical data input in step S22 is different from the medical data input in step S21. Step S22 may be omitted.

In step S23, the learning function 254 builds a trained model for estimating information on health conditions based on medical data. The medical data input in step S23 is the same as the medical data input in step S21.

Note that steps S21-23 may be executed in random order.

Through the above procedure, a trained model and a simulation model can be constructed for each estimation process.

FIG. 6 is a flow chart showing an example of the procedure when the estimation process is operated by the simulation model and part of the input medical data required for the estimation process is insufficient. In the estimation process shown in FIG. 6, it is assumed that operation by a simulation model has already been determined as a result of optimization.

In step S<b>31 , the acquisition function 251 acquires the medical data of the subject to be estimated from the subject DB 32 .

Next, in step S32, the estimation function 252 determines whether or not there is a shortage of medical data necessary for estimation processing using the simulation model. If there is no shortage, the process proceeds to step S36. On the other hand, if there is a shortage, the process proceeds to step S33 to determine whether or not the missing data can be measured.

If measurable, the process proceeds to step S34, the missing medical data is collected by measuring the subject, and stored in the subject DB 32 in association with the subject, and the process proceeds to step S36. On the other hand, if the measurement is not possible, the process proceeds to step S35, the missing medical data is estimated by the trained model for estimating the missing medical data, and the process proceeds to step S36. In this case, the medical data estimated by the trained model should be stored in the subject DB 32 in association with the subject by adding information indicating that it is data that has not been measured (unmeasured data). Next, when the same medical data is insufficient when executing the procedure shown in FIG. 6, the non-measurement data stored in the subject DB 32 can be used without estimating the learned model.

Next, in step S36, the estimation function 252 estimates information about the health condition of the subject to be estimated using a simulation model.

Then, in step S<b>37 , the notification function 255 notifies at least one of the user and the subject of the estimation result of the information regarding the health condition of the subject estimated by the estimation function 252 .

According to the above procedure, when the simulation model is used for the estimation process and part of the input medical data required for the estimation process is insufficient, the learned model can be used to supplement the missing data. Therefore, a highly accurate estimation result can be obtained more easily and efficiently than when using only a simulation model.

Next, specific examples of model construction and model operation will be described with reference to FIGS. 7 and 8, respectively.

FIG. 7 is a flowchart showing an example of a specific procedure for constructing a learned model and a simulation model for estimating coronary artery blood flow parameters.

In step S210, the learning function 254 constructs a blood flow parameter simulation model X for estimating blood flow parameters based on boundary conditions, blood flow viscosity, and X-ray CT image data.

In step S221, the learning function 254 builds a trained model A that estimates boundary conditions based on blood pressure and heart volume.

In step S222, the learning function 254 builds a trained model B for estimating blood flow viscosity based on blood test results.

In step S230, the learning function 254 constructs a blood flow parameter trained model C for estimating blood flow parameters based on boundary conditions, blood flow viscosity, and X-ray CT image data.

Note that these steps S210-230 may be executed in any order.

Through the above procedure, a trained model and a simulation model can be constructed for the process of estimating blood flow parameters of coronary arteries.

FIG. 8 is a flow chart showing an example of the procedure when the simulation model X is used to estimate the blood flow parameters of the coronary arteries and part of the input medical data required for the estimation process is insufficient. In the estimation process shown in FIG. 8, as in the example shown in FIG. 6, it is assumed that the simulation model operation has already been determined as a result of the optimization.

In step S<b>310 , the acquisition function 251 acquires the medical data of the subject to be estimated from the subject DB 32 .

Next, in step S320, the estimating function 252 determines whether or not the boundary conditions necessary for estimating the coronary artery blood flow parameters by the simulation model X (see step S210 in FIG. 7) are sufficient. If there is no shortage, the process proceeds to step S360. On the other hand, if there is a shortage, the flow advances to step S330 to determine whether or not the missing boundary conditions are measurable.

If it is measurable, the process proceeds to step S340, the object is measured and collected with respect to the missing boundary conditions, the data is stored in the object DB 32 in association with the object, and the process proceeds to step S360. On the other hand, if the measurement is not possible, the flow advances to step S350 to estimate the missing boundary conditions using the learned model B (see step S222 in FIG. 7) for estimating the missing boundary conditions, and then to step S360. In this case, the boundary conditions estimated by the trained model B should be stored in the subject DB 32 in association with the subject by adding information indicating that they are unmeasured data (unmeasured data).

Next, in step S360, the estimation function 252 estimates the blood flow parameter of the coronary arteries of the subject to be estimated using the simulation model X. FIG.

Then, in step S370, the notification function 255 notifies at least one of the user and the subject of the estimation result of the coronary artery blood flow parameter estimated by the estimation function 252. FIG.

According to the above procedure, when the simulation model X is used to estimate the coronary artery blood flow parameters and the boundary conditions are insufficient, the learned model B can be used to estimate and complement the boundary conditions. Thus, the medical information processing apparatus 20 according to this embodiment can flexibly optimize the combination of the simulation model and the trained model.

FIG. 9 is a flowchart showing an example of a procedure when estimation processing is operated using a trained model and part of the input medical data necessary for estimation processing is insufficient. In the estimation process shown in FIG. 8, contrary to the example shown in FIG. 6, it is assumed that operation by the learned model has already been determined as a result of optimization.

In step S<b>41 , the acquisition function 251 acquires the medical data of the subject to be estimated from the subject DB 32 .

Next, in step S42, the estimation function 252 determines whether or not there is a shortage of medical data necessary for estimation processing using the trained model. If there is no shortage, the process proceeds to step S46. On the other hand, if there is a shortage, the process proceeds to step S43 to determine whether or not the missing data can be measured.

If it is measurable, the process proceeds to step S44, the missing medical data is collected by measuring the subject, associated with the subject and stored in the subject DB 32, and the process proceeds to step S46. On the other hand, if the measurement is not possible, the process proceeds to step S45 to estimate the missing medical data by a simulation model for estimating the missing medical data, and the process proceeds to step S46. In this case, the medical data estimated by the trained model should be stored in the subject DB 32 in association with the subject by adding information indicating that it is data that has not been measured (unmeasured data).

Next, in step S46, the estimation function 252 estimates information about the health condition of the subject to be estimated using the learned model.

Then, in step S<b>47 , the notification function 255 notifies at least one of the user and the subject of the estimation result of the information regarding the health condition of the subject estimated by the estimation function 252 .

According to the above procedure, when the estimation process is operated by the trained model and part of the input medical data necessary for the estimation process is insufficient, the missing data can be supplemented using the simulation model.

The medical information processing apparatus 20 according to this embodiment can determine which of the trained model and the simulation model to use for each estimation process. Whether to use the trained model or the simulation model can be determined by the optimization function 253 based on at least one of the degree of matching between each estimation result and the correct estimation result, and the time required for each estimation. Whether to use the degree of matching or time, or to use both can be set by the user. Therefore, the user can improve the accuracy of the estimation result or improve the estimation processing speed by appropriately combining the models according to the purpose of use.

According to at least one embodiment described above, the health condition of the subject can be estimated by combining the estimation by the trained model and the estimation by the simulation model.

In the above-described embodiments, the term “processor” means, for example, a dedicated or general-purpose CPU (Central Processing Unit), a GPU (Graphics Processing Unit), or an application-specific integrated circuit (ASIC), a programmable logic device (for example, a Simple Programmable Logic Device (SPLD), a Complex Programmable Logic Device (CPLD), and an FPGA) or the like. The processor implements various functions by reading and executing programs stored in the storage medium.

In the above embodiments, an example in which a single processor of the processing circuit implements each function has been shown, but a processing circuit may be configured by combining a plurality of independent processors, and each processor may implement each function. Further, when a plurality of processors are provided, a storage medium for storing programs may be provided individually for each processor, or one storage medium may collectively store programs corresponding to the functions of all processors.

It should be noted that although several embodiments of the invention have been described, these embodiments are provided by way of example and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, replacements, and modifications can be made without departing from the scope of the invention. These embodiments and their modifications are included in the scope and spirit of the invention, as well as the scope of the invention described in the claims and equivalents thereof.

10 medical information processing system 20 medical information processing apparatus 51 wearable terminal 100 network 251 acquisition function 252 estimation function 253 optimization function 254 learning function 255 notification function 451 estimation request function 452 estimation result acquisition function 453 image generation function

Claims (13)

an acquisition unit that acquires a correct data set from a storage unit that stores a correct data set consisting of a set of medical data for verification collected from a plurality of subjects and correct information about health conditions;
an optimization unit that determines which of the trained model and the simulation model is used to estimate the information on the health condition of the subject to be estimated, based on an estimation result of information on the health condition generated by the learned model based on the medical data for verification, an estimation result of the information on the health condition generated by the simulation model based on the medical data for verification, and a correct answer of the information on the health condition;
medical information processing device. The estimation of information about the health condition of the subject to be estimated includes:
at least one estimation process of estimating a current medical parameter of the estimation target subject, estimating a current health status of the estimation target subject, estimating a future health status including estimating a future prognosis of the estimation target subject, and estimating a treatment plan for the estimation target subject;
The trained model and the simulation model are
preconfigured for each of the at least one estimation process;
The optimization unit
determining which of the trained model and the simulation model is to be used for each of the at least one estimation process;
The medical information processing apparatus according to claim 1. The at least one estimation process includes:
It consists of multiple estimation processes,
The trained model and the simulation model are
Pre-configured for each of the plurality of estimation processes,
The optimization unit
Determining which of the trained model and the simulation model is used for estimation for each of the plurality of estimation processes;
The medical information processing apparatus according to claim 2. The optimization unit
Based on the information of the plurality of estimation processes, inputting the information of the plurality of estimation processes to a combined estimation trained model that generates a combination of models responsible for each of the plurality of estimation processes, thereby determining the combination of models responsible for each of the plurality of estimation processes.
The medical information processing apparatus according to claim 3. a learning unit that builds in advance the trained model and the simulation model for each of the estimation processes;
The medical information processing apparatus according to any one of claims 2 to 4, further comprising: The acquisition unit
obtaining medical data of the subject to be estimated;
an estimating unit that estimates information related to the health condition of the subject to be estimated based on the medical data of the subject to be estimated using either the trained model determined by the optimization unit or the simulation model;
further comprising
The medical information processing apparatus according to any one of claims 1 to 5. The estimation unit
when the medical data of the subject to be estimated is insufficient, supplementing the medical data by estimating using the other of the trained model determined by the optimization unit and the simulation model;
The medical information processing apparatus according to claim 6. When the optimization unit determines to estimate information about the health condition of the subject to be estimated using the simulation model, and the medical data of the subject to be estimated used for estimation using the simulation model is insufficient, estimate and complement the missing medical data using the trained model.
The medical information processing apparatus according to claim 7. a notification unit that notifies at least one of a user and the estimation target subject of an estimation result of information related to the health condition of the estimation target subject estimated by the estimation unit;
The medical information processing apparatus according to any one of claims 6 to 8, further comprising: The optimization unit
further based on the time required for each of the trained model and the simulation model to generate an estimation result, determining which of the trained model and the simulation model is used to estimate information about the health condition of the subject to be estimated;
The medical information processing apparatus according to any one of claims 1 to 9. A medical information processing system, comprising: a medical information processing apparatus for estimating information about a health condition of a subject using at least one of a trained model and a simulation model based on medical data of the estimation target subject collected from the estimation target subject; and a client capable of communicating with the medical information processing apparatus via a network,
The medical information processing device is
an acquisition unit that acquires a correct data set from a storage unit that stores a correct data set consisting of a set of medical data for verification collected from a plurality of subjects and correct information about health conditions;
an optimization unit that determines which of the trained model and the simulation model is used to estimate information about the health condition of the subject to be estimated, based on an estimation result of information about the health condition generated by the learned model based on the medical data for verification, an estimation result of the information about the health condition generated by the simulation model based on the medical data for verification, and a correct answer of the information about the health condition;
an estimating unit that estimates information related to the health condition of the subject to be estimated based on the medical data of the subject to be estimated using either the trained model determined by the optimization unit or the simulation model;
a transmission unit configured to transmit, to the client, an estimation result of information related to the health condition of the subject to be estimated by the estimation unit;
medical information processing system. Said client
a receiving unit that receives, from the medical information processing apparatus, an estimation result of information about the health condition of the subject to be estimated by the estimating unit;
an image generation unit configured to generate an image for supporting diagnosis of the subject based on the estimation result received by the estimation unit and to display the image on a display unit;
The medical information processing system according to claim 11, comprising: A medical information processing method executed by a medical information processing apparatus comprising an acquisition unit, an estimation unit, and an optimization unit,
a step of acquiring the correct data set from a storage unit storing a correct data set consisting of a set of verification medical data collected from a plurality of subjects by the acquiring unit and a result of estimating the correctness of information related to health conditions;
a step in which the estimating unit generates an estimation result of information about the health condition of the trained model based on the verification medical data;
a step in which the estimation unit generates an estimation result of information about the health condition by the simulation model based on the verification medical data;
determining , by the optimization unit , which of the trained model and the simulation model is used to estimate the information regarding the health condition of the subject to be estimated, based on the estimation result of the trained model, the estimation result of the simulation model, and the correct answer of the information regarding the health condition;
A medical information processing method comprising:

Images