Patent Application
20230046986
Embodiments disclosed in the present description and drawings relate to a medical information processing system, a medical information processing method, and a storage medium.
Although the replies of a patient in a medical examination by interview are important information for estimating the condition of the patient, this is unstable information that often fluctuates due to the patient’s own sensitivity and feelings. Currently, medical staff assimilate and understand these fluctuations and use it for medical treatment. Meanwhile, automation of medical treatment using artificial intelligence (AI) is under review. However, instability of replies of a patient in medical examinations by interview may have undesirable effects on medical treatment using AI.
Hereinafter, a medical information processing system, a medical information processing method, and a storage medium of embodiments will be described with reference to the drawings.
A medical information processing system of an embodiment includes a processing circuit. The processing circuit acquires examination data showing medical examination results with respect to a medical treatment subject and reply data showing reply results in medical examinations by interview with respect to the medical treatment subject. The processing circuit estimates information on medical treatment of the medical treatment subject by inputting the examination data of the medical treatment subject to a first model. The first model is a model trained on the basis of a first training data set in which information on medical treatment of a learning subject is associated with the examination data of the learning subject as a correct label. Further, the processing circuit estimates information on medical treatment of the medical treatment subject by inputting the examination data and the reply data of the medical treatment subject to a second model. The second model is a model trained on the basis of a second training data set in which the information on medical treatment of the learning subject is associated with the examination data and the reply data of the learning subject as a correct label. The processing circuit outputs a first estimation result representing the information on medical treatment estimated using the first model and a second estimation result representing the information on medical treatment estimated using the second model via an output unit.
The communication network NW may mean an information communication network in general using telecommunications technology. For example, the communication network NW includes a telephone communication line network, an optical fiber communication network, a cable communication network, a satellite communication network, and the like in addition to a wireless/wired local area network (LAN) such as a hospital backbone LAN and the Internet.
The user interface 10 is used by patients and medical staff. For example, the user interface 10 is a touch interface or a voice user interface, and more specifically, a terminal device such as a personal computer, a tablet terminal, or a mobile phone. Medical staff are typically doctors but may be nurses or other people involved in medical treatment. For example, a patient inputs his/her own replies in a medical examination by interview to the user interface 10 through touch or voice input. A medical staff member may perform an oral medical examination by interview for the patient, listens to replies from the patient, and input listening comprehension results to the user interface 10.
In the present embodiment, “medical treatment” may include not only treatment such as surgery and medication but also examination up to treatment or after treatment and any other medical practices up to or after the treatment.
The user interface 10 transmits information input by a patient or a medical staff member to the medical information processing device 100 via the communication network NW or receives information from the medical information processing device 100.
The medical information processing device 100 receives information from the user interface 10 via the communication network NW and processes the received information. Then, the medical information processing device 100 transmits the processed information to the user interface 10 via the communication network NW. In addition to or instead of transmitting the processed information to the user interface 10, the medical information processing device 100 may transmit the processed information to a dedicated terminal of a medical staff member installed in a hospital.
The medical information processing device 100 may be a single device or may be a system in which a plurality of devices connected via the communication network NW operate in cooperation. That is, the medical information processing device 100 may be realized by a plurality of computers (processors) included in a distributed computing system or a cloud computing system. The medical information processing device 100 does not necessarily have to be a separate device from the user interface 10 and may be a device integrated with the user interface 10.
The communication interface 11 communicates with the medical information processing device 100 or the like via the communication network NW. The communication interface 11 includes, for example, a network interface card (NIC), an antenna for wireless communication, and the like.
The input interface 12 receives various input operations from an operator (for example, a patient), converts the received input operations into electrical signals, and outputs the electrical signals to the processing circuit 20. For example, the input interface 12 includes a mouse, a keyboard, a trackball, a switch, a button, a joystick, a touch panel, and the like. The input interface 12 may be, for example, a user interface that receives voice input, such as a microphone. When the input interface 12 is a touch panel, the input interface 12 may also have a display function of a display 13a included in the output interface 13 which will be described later.
In the present description, the input interface 12 is not limited to one including physical operation parts such as a mouse and a keyboard. For example, examples of the input interface 12 include an electrical signal processing circuit that receives an electrical signal corresponding to an input operation from an external input device provided separately from the device and outputs the electrical signal to a control circuit.
The output interface 13 includes, for example, the display 13a, a speaker 13b, and the like.
The display 13a displays various types of information. For example, the display 13a displays an image generated by the processing circuit 20, a graphical user interface (GUI) for receiving various input operations from an operator, and the like. For example, the display 13a is a liquid crystal display (LCD), a cathode ray tube (CRT) display, an organic electroluminescence (EL) display, or the like.
The speaker 13b outputs information input from the processing circuit 20 as voice information.
The memory 14 is realized by, for example, a semiconductor memory element such as a random access memory (RAM) or a flash memory, a hard disk, or an optical disc. These non-transient storage media may be realized by other storage devices connected via the communication network NW, such as a network attached storage (NAS) and an external storage server device. The memory 14 may include a non-transitory storage medium such as a read only memory (ROM) or a register.
The processing circuit 20 includes, for example, an acquisition function 21, an output control function 22, and a communication control function 23. The processing circuit 20 realizes these functions by, for example, a hardware processor (computer) executing a program stored in the memory 14 (storage circuit).
The hardware processor in the processing circuit 20 is defined as, for example, a circuit (circuitry) such as a central processing unit (CPU), a graphics processing unit (GPU), an application specific integrated circuit (ASIC), or a programmable logic device (for example, a simple programmable logic device (SPLD) or a complex programmable logic device (CPLD), or a field programmable gate array (FPGA)). Instead of storing the program in the memory 14, the program may be configured to be directly embedded in the circuit of the hardware processor. In this case, the hardware processor realizes the functions by reading and executing the program embedded in the circuit. The aforementioned program may be stored in the memory 14 in advance, or may be stored in a non-temporary storage medium such as a DVD or a CD-ROM and installed in the memory 14 from the non-temporary storage medium when the non-temporary storage medium is set in a drive device (not shown) of the user interface 10. The hardware processor is not limited to the one configured as a single circuit, and may be configured as one hardware processor by combining a plurality of independent circuits to realize each function. A plurality of components may be integrated into one hardware processor to realize each function.
The acquisition function 21 acquires input information via the input interface 12 or acquires information from the medical information processing device 100 via the communication interface 11.
The output control function 22 displays information acquired by the acquisition function 21 on the display 13a or outputs the information through the speaker 13b.
The communication control function 23 transmits information input to the input interface 12 to the medical information processing device 100 via the communication interface 11.
The communication interface 111 communicates with the user interface 10 and the like via the communication network NW. The communication interface 111 includes, for example, an NIC or the like. The communication interface 111 is an example of an “output unit.”
The input interface 112 receives various input operations from an operator, converts the received input operations into electrical signals, and outputs the electrical signals to the processing circuit 120. For example, the input interface 112 includes a mouse, a keyboard, a trackball, a switch, a button, a joystick, a touch panel, and the like. The input interface 112 may be, for example, a user interface that receives voice input, such as a microphone. When the input interface 112 is a touch panel, the input interface 112 may also have a display function of a display 113a included in the output interface 113 which will be described later.
In the present description, the input interface 112 is not limited to one provided with physical operating parts such as a mouse and a keyboard. For example, examples of the input interface 112 include an electrical signal processing circuit that receives an electrical signal corresponding to an input operation from an external input device provided separately from the device and outputs the electrical signal to a control circuit.
The output interface 113 includes, for example, the display 113a, a speaker 113b, and the like. The output interface 113 is another example of the “output unit”.
The display 113a displays various types of information. For example, the display 113a displays an image generated by the processing circuit 120, a GUI for receiving various input operations from the operator, and the like. For example, the display 113a is an LCD, a CRT display, an organic EL display, or the like.
The speaker 113b outputs information input from the processing circuit 120 as voice information.
The memory 114 is realized by, for example, a semiconductor memory element such as a RAM or a flash memory, a hard disk, or an optical disc. These non-transient storage media may be realized by other storage devices connected via a communication network NW, such as a NAS or an external storage server device. The memory 114 may include a non-transitory storage medium such as a ROM or a register.
The memory 114 stores model information in addition to a program executed by a hardware processor. The model information is information (program or data structure) that defines a first model MDL1 and a second model MDL2. The first model MDL1 and the second model MDL2 may be implemented by a deep neural network (DNN) such as a convolutional neural network (CNN), for example. The first model MDL1 and the second model MDL2 are not limited to a DNN and may be implemented by other models such as a support vector machine, a decision tree, a naive Bayes classifier, and a random forest. Details of the first model MDL1 and the second model MDL2 will be described later.
When the first model MDL1 and the second model MDL2 are implemented by a DNN, the model information includes, for example, coupling information on how units included an input layer constituting the DNN, one or more hidden layers (intermediate layers), and an output layer are coupled with each other, and weight information on how many coupling coefficients are provided to data input/output between coupled units. The coupling information is, for example, information for specifying the number of units included in each layer, information for specifying the type of a unit that is a coupling destination of each unit, an activation function for realizing each unit, and a gate provided between units in a hidden layer. The activation function for realizing a unit may be, for example, a rectified linear unit (ReLU) function, an exponential linear units (ELU) function, a clipping function, a Sigmoid function, a step function, a hyperpolic tangent function, an equality function, or the like. The gate selectively passes or weights data transmitted between units, for example, depending on a value (e.g., 1 or 0) returned by the activation function. The coupling coefficients include, for example, a weight applied to output data when data is output from a unit of a certain layer to a unit of a deeper layer in a hidden layer of a neural network. The coupling coefficients may include a bias component specific to each layer, and the like.
The processing circuit 120 includes, for example, an acquisition function 121, an estimation function 122, a determination function 123, an output control function 124, and a communication control function 125. The acquisition function 121 is an example of an “acquisition unit,” the estimation function 122 is an example of an “estimation unit,” and the determination function 123 is an example of a “determination unit.” The output control function 124 is an example of an “output control unit” and the communication control function 125 is another example of the “output control unit.”
The processing circuit 120 realizes these functions by, for example, a hardware processor (computer) executing a program stored in the memory 114 (storage circuit).
The hardware processor in the processing circuit 120 means, for example, a circuit (circuitry) such as a CPU, a GPU, an application specific integrated circuit, or a programmable logic device (for example, a simple programmable logic device or a complex programmable logic device, or a field programmable gate array). Instead of storing the program in the memory 114, the program may be configured to be directly embedded in the circuit of the hardware processor. In this case, the hardware processor realizes the functions by reading and executing the program embedded in the circuit. The aforementioned program may be stored in the memory 114 in advance or may be stored in a non-temporary storage medium such as a DVD or a CD-ROM and installed in the memory 114 from the non-temporary storage medium when the non-temporary storage medium is set in a drive device (not shown) of the medical information processing device 100. The hardware processor is not limited to one configured as a single circuit and may be configured as one hardware processor by combining a plurality of independent circuits to realize each function. A plurality of components may be integrated into one hardware processor to realize each function.
Hereinafter, a series of processing performed by the processing circuit 120 of the medical information processing device 100 will be described with reference to a flowchart.
First, the acquisition function 121 acquires reply data of a patient that is a medical treatment target (hereinafter, a medical treatment subject) for a medical examination by interview from the user interface 10 via the communication interface 111 and further acquires examination data of the medical treatment subject from examination equipment (step S100).
The examination equipment is equipment for medically examining patients and is, for example, an X-ray computed tomography (CT) device, a magnetic resonance imaging (MRI) device, a mammography device, a sound imaging diagnostic device, a nuclear medicine diagnostic device, a body fluid analysis device, devices for measuring vital signs, or the like. Examination data is quantitative digital data obtained by measuring biological information of the medical treatment subject by the above-mentioned various types of examination equipment. On the other hand, reply data is qualitative digital data including the subjectivity of the medical treatment subject. That is, the acquisition function 121 acquires quantitative data and qualitative data regarding the medical treatment subject.
Return to description of the flowchart of
Information on medical treatment is, for example, information on medical treatment that should be taken after a starting point when a time at which the learning subject undergoes an examination is assumed as the starting point. Information on medical treatment is typically, but not limited to, estimating a disease that the patient has already suffered at the starting point or a disease that the patient is highly likely to suffer in the future. For example, information on medical treatment may include estimating or determining the name and type of an examination, prescription drug name, treatment policy, whether a patient can return home, the type of a patient’s room, presence or absence of help of other staff, and the like. That is, information on medical treatment may include all matters in the future determined by medical examinations by interview. In the following description, as an example, it is assumed that information on medical treatment is “disease estimation.”
When the information on medical treatment is “disease estimation,” training data for learning the first model MDL1 is a data set in which a disease that a learning subject has already suffered from or a disease that the learning subject is highly likely to suffer in the future is associated with examination data of the learning subject as a correct label.
When examination data of a certain patient is input, as illustrated, the first model MDL1 trained using such training data outputs, as an estimation result, the disease of the patient (an example of “information on medical treatment”). The estimation result of the first model MDL1 is represented by, for example, a multidimensional vector or tensor. The vector or tensor includes the probability of being a disease as an element value. For example, it is assumed that there are a total of three types of diseases that the medical treatment subjects can suffer: disease A, disease B, and disease C. In this case, the vector or tensor can be represented as (e1, e2, e3) where the probability of disease A is e1, the probability of disease B is e2, and the probability of disease C is e3.
Return to description of the flowchart in
On the other hand, when the reply data and the examination data for the medical examination by interview are acquired by the acquisition function 121, the estimation function 122 inputs the reply data and the examination data of the medical treatment subject acquired by the acquisition function 121 to the second model MDL2 defined by the model information stored in the memory 141 (step S106).
Information on medical treatment here is, for example, information on medical treatment that should be taken after a starting point when a time at which a learning subject undergoes an examination or a time at which the learning subject replies to a medical examination by interview is assumed as the starting point. As described above, information on medical treatment is typically, but not limited to, estimating a disease that the patient has already suffered at the starting point or a disease that the patient is highly likely to suffer in the future, and may include all matters in the future determined by medical examinations by interview. In the following description, as an example, it is assumed that information on medical treatment is “disease estimation.”
When information on medical treatment is “disease estimation,” training data for learning the second model MDL2 is a data set in which a disease that a learning subject has already suffered or a disease that the learning subject is highly likely to suffer in the future is associated with reply data and examination data of the learning subject as a correct label.
When reply data and examination data of a certain patient are input, as illustrated, the second model MDL2 trained using such training data outputs, as an estimation result, the disease of the patient. The estimation result of the second model MDL2 may be represented by a multidimensional vector or tensor like the estimation result of the first model MDL1.
Return to description of the flowchart in
Next, the output control function 124 outputs a first estimation result representing the disease of the medical treatment subject estimated using the first model MDL1 and a second estimation result representing the disease of the medical treatment subject estimated using the second model MDL2 via the output interface 113 (step S110). Accordingly, processing of this flowchart ends.
The communication control function 125 may transmit the first estimation result and the second estimation result to the user interface 10 via the communication interface 111. When the communication interface 11 receives the first estimation result and the second estimation result from the medical information processing device 100, the output control function 22 of the user interface 10 causes the display 13a of the output interface 13 to display the estimation results as an image or causes the estimation results to be output as voice through the speaker 13b.
According to the first embodiment described above, the processing circuit 120 of the medical information processing device 100 acquires examination data of a medical treatment subject and reply data for a medical examination by interview. The processing circuit 120 estimates information on medical treatment of the medical treatment subject by inputting the examination data of the medical treatment subject to the first model MDL1 trained in advance. For example, the processing circuit 120 may estimate the disease of the medical treatment subject as the information on medical treatment.
The processing circuit 120 estimates information on medical treatment of the medical treatment subject by inputting the examination data and the reply data of the medical treatment subject to the second model MDL2 trained in advance. For example, the processing circuit 120 may estimate the disease of the medical treatment subject as the information on medical treatment.
Then, the processing circuit 120 outputs the first estimation result representing the information on medical treatment (e.g., the disease of the medical treatment subject) estimated using the first model MDL1 and the second estimation result representing the information on medical treatment (e.g., the disease of the medical treatment subject) estimated using the second model MDL2 via the output interface 113. Accordingly, a medical staff member can treat the patient without being affected by instability of the replies of the patient in a medical examination by interview.
Hereinafter, a second embodiment will be described. The second embodiment differs from the first embodiment in that it is determined whether or not the first estimation result and the second estimation result match. Hereinafter, differences from the first embodiment will be mainly described, and the points common to the first embodiment will be omitted. In the description of the second embodiment, the same parts as those of the first embodiment will be described with the same reference numerals.
The determination function 123 of the second embodiment compares the first estimation result representing information on medical treatment (for example, the disease of the medical treatment subject) estimated using the first model MDL1 to the second estimation result representing information on medical treatment (for example, the disease of the medical treatment subject) estimated using the second model MDL2 and determines whether or not the estimation results match.
When the determination function 123 determines that the first estimation result and the second estimation result do not match, the output control function 124 of the second embodiment outputs an alert AR via the output interface 113 in order to notify a medical staff member that the estimation results do not match.
According to the second embodiment described above, the processing circuit 120 compares the first estimation result and the second estimation result and determines whether or not the first estimation result and the second estimation result match. Upon determining that the first estimation result and the second estimation result do not match, the processing circuit 120 outputs the alert AR via the output interface 113. Accordingly, it is possible to more strongly remind a medical staff member that instability of reply data can affect disease estimation results.
Hereinafter, a third embodiment will be described. The third embodiment differs from the above-described embodiments in that, when information on medical treatment of a medical treatment subject is estimated by inputting examination data and reply data of the medical treatment subject to the second model MDL2 trained in advance, the reply data is weighted. Hereinafter, differences from the first embodiment and the second embodiment will be mainly described, and the points common to the first embodiment and the second embodiment will be omitted. In the description of the third embodiment, the same parts as those of the first embodiment or the second embodiment will be described with the same reference numerals.
The estimation function 122 repeats inputting the reply data to the second model MDL2 while changing the weighting factor. As a result, the second estimation result is obtained for each weighting factor.
The determination function 123 of the third embodiment compares a plurality of second estimation results obtained for respective weighting factors with each other and determines whether or not the second estimation results match each other.
The output control function 124 of the third embodiment displays, for example, a plurality of second estimation results obtained for respective weighting factors side by side on the display 113a. When the determination function 123 determines that the second estimation results do not match each other, the output control function 124 may also display the alert AR on the display 113a.
According to the third embodiment described above, the processing circuit 120 weight reply data at the time of estimating information on medical treatment of a medical treatment subject by inputting examination data and the reply data of the medical treatment subject to the second model MDL2 trained in advance. The processing circuit 120 repeats estimating information on medical treatment while changing the weighting factor. The processing circuit 120 compares second estimation results respectively representing pieces of information on medical treatment repeatedly estimated using the second model MDL2 and determines whether or not the plurality of second estimation results match each other. Then, the processing circuit 120 outputs the second estimation result for each weighting factor. Accordingly, for example, if the second estimation result does not change even if the weighting factor is changed, it can be determined that the reply data of the medical treatment subject does not affect the estimation result of the information on medical treatment. That is, it is possible to accurately estimate the information on medical treatment even if the reply data of the medical treatment subject is used.
Hereinafter, a fourth embodiment will be described. The fourth embodiment differs from the above-described embodiments in that a weighting factor of reply data is determined on the basis of the age of a medical treatment subject, a period elapsed since the medical treatment subject suffered from a disease (onset period), the experience of replying of the medical treatment subject in a medical examination by interview, and the like. Hereinafter, differences from the first to third embodiments will be mainly described, and the points common to the first to third embodiments will be omitted. In description of the fourth embodiment, the same parts as those of the first to third embodiments will be described with the same reference numerals.
As in the third embodiment, the estimation function 122 of the fourth embodiment weights reply data of a medical treatment subject at the time of inputting the reply data of the medical treatment subject to the second model MDL2 trained in advance. At this time, the estimation function 122 of the fourth embodiment determines a weighting factor of the reply data on the basis of the age of the medical treatment subject, a period elapsed since the medical treatment subject suffered from a disease (onset period), the experience of replying of the medical treatment subject in a medical examination by interview, and the like.
For example, children and the elderly are more likely to have fluctuations in replying to medical examinations by interview than adults. Therefore, the estimation function 122 in the fourth embodiment may decrease the weighting factor of reply data as a medical treatment subject becomes younger than a certain reference age (for example, 18 years old). Similarly, the estimation function 122 may decrease the weighting factor of reply data as the medical treatment subject becomes older than a certain reference age (for example, 65 years).
A patient who has just suffered from a disease is more likely to have more fluctuations in replying to medical examinations by interview than patients who do not. Therefore, the estimation function 122 in the fourth embodiment may decrease the weighting factor of reply data as a period elapsed since a medical treatment subject suffered from a disease is shorter (shorter after suffering from the disease).
A patient who replies to a medical examination by interview for the first time at the first visit is more likely to have more fluctuations in replying to the medical examination by interview than experienced patients who have already been examined many times and replied to medical examinations by interviews. Therefore, the estimation function 122 in the fourth embodiment may decrease the weighting factor of reply data as the medical treatment subject has less experience in replying to medical examinations by interview.
Accordingly, it is possible to remove the influence of fluctuations in replies caused by the age, onset period, reply experience, and the like of a medical treatment subject from the output result of a trained model MDL. As a result, a medical staff member can further treat patients while considering the uncertainty of a trained model MDL.
Hereinafter, a fifth embodiment will be described. The fifth embodiment differs from the above-described embodiments in that a plurality of machine learning models are provided for ages of medical treatment subjects, a plurality of machine learning models are provided for onset periods of medical treatment subjects, and a plurality of machine learning models are provided for experiences of replying of medical treatment subjects in medical examinations by interview. Hereinafter, differences from the first to fourth embodiments will be mainly described, and the points common to the first to fourth embodiments will be omitted. In description of the fifth embodiment, the same parts as those of the first to fourth embodiments will be described with the same reference numerals.
For example, the processing circuit 120 may generate a first model MDL1 for children by using training data in which information on medical treatment (for example, a disease) of a child under the age of 18 is associated with examination data of the child as a correct label, and generate a second model MDL2 for children by using training data in which the information on medical treatment (for example, a disease) of the child under the age of 18 is associated with reply data and examination data of the child as a correct label.
Similarly, the processing circuit 120 may generate a first model MDL1 for adults by using training data in which information on medical treatment (for example, a disease) of an adult between ages of 18 and 65 is associated with examination data of the adult as a correct label, and generate a second model MDL2 for adults by using training data in which the information on medical treatment (for example, a disease) of the adult between ages of 18 and 65 is associated with reply data and examination data of the adult as a correct label.
The processing circuit 120 may generate a first model MDL1 for the elderly by using training data in which information on medical treatment (for example, a disease) of the elderly over the age of 65 is associated with examination data of the elderly as a correct label, and generate a second model MDL2 for the elderly by using training data in which the information on medical treatment (for example, a disease) of the elderly over the age of 65 is associated with reply data and examination data of the elderly as a correct label. Accordingly, it is possible to remove the influence of reply fluctuations due to an age from the uncertainty of a trained model MDL.
For example, the processing circuit 120 of the fifth embodiment may generate a first model MDL1 specialized for patients having onset periods of less than 1 year by using training data in which information on medical treatment (for example, a disease) of a patient having an onset period of less than 1 year is associated with examination data of the patient as a correct label, and generate a second model MDL2 specialized for patients having onset periods of less than 1 year by using training data in which information on medical treatment (for example, a disease) of a patient having an onset period of less than 1 year is associated with reply data and examination data of the patient as a correct label.
The processing circuit 120 may generate a first model MDL1 specialized for patients having onset periods of 1 year or longer by using training data in which information on medical treatment (for example, a disease) of a patient having an onset period of 1 year or longer is associated with examination data of the patient as a correct label, and generate a second model MDL2 specialized for patients having onset periods of 1 year or longer by using training data in which information on medical treatment (for example, a disease) of a patient having an onset period of 1 year or longer is associated with reply data and examination data of the patient as a correct label. Accordingly, it is possible to remove the influence of reply fluctuations due to an onset period from the uncertainty of a trained model MDL.
For example, the processing circuit 120 of the fifth embodiment may generate a first model MDL1 specialized for first-visit patients by using training data in which information on medical treatment (for example, a disease) of a first-visit patient having no experience of replying to medical examinations by interview is associated with examination data of the patient as a correct label, and generate a second model MDL2 specialized for first-visit patients by using training data in which the information on medical treatment (for example, a disease) of the first-visit patient is associated with reply data and examination data of the first-visit patient as a correct label.
For example, the processing circuit 120 of the fifth embodiment may generate a first model MDL1 specialized for patients who attend second and subsequent medical examinations by using training data in which information on medical treatment (for example, a disease) of a patient who has one or more experiences of replying to medical examinations by interview in second and subsequent medical examinations is associated with examination data of the patient as a correct label, and generate a second model MDL2 specialized for patients who attend second and subsequent medical examinations by using training data in which the information on medical treatment (for example, a disease) of the patient who has one or more experiences of replying to medical examinations by interview in second and subsequent medical examinations is associated with reply data and examination data of the patient as a correct label. Accordingly, it is possible to remove the influence of reply fluctuations due to experience of medical examination by interview from the uncertainty of a trained model MDL.
Hereinafter, other embodiments will be described. Although the first model MDL1 and the second model MDL2 have been described as different models in the above-described embodiments, the present invention is not limited thereto and they may be the same model. The first model MDL1 and the second model MDL2 may have some parameters (such as DNN weights and bias components) different from each other and the remaining parameters that are common.
Although the user interface 10 and the medical information processing device 100 have been described as different devices in the above-described embodiments, the present invention is not limited thereto. For example, the user interface 10 and the medical information processing device 100 may be a single integrated device. For example, the processing circuit 20 of the user interface 10 may further include the estimation function 122 and a determination function 123 included in the processing circuit 120 of the medical information processing device 100 in addition to the acquisition function 21, the output control function 22, and the communication control function 23. In this case, the user interface 10 can perform various types of processing of the flowcharts described above standalone (offline).
As also described in the above-described embodiments, information on medical treatment output by the first model MDL1 and the second model MDL2 is typically estimating a disease of a patient, but is not limited thereto and may include all matters in the future determined by medical examinations by interview, such as the name and type of an examination, the name of prescription drug, treatment policy, whether or not a patient can return home, the type of a patient’s room, and whether or not there is help from other staff.
According to at least one embodiment described above, the processing circuit 120 of the medical information processing device 100 acquires examination data and reply data of a medical examination by interview of a medical treatment subject. The processing circuit 120 estimates information on medical treatment of the medical treatment subject by inputting the examination data of the medical treatment subject to the first model MDL1 trained in advance. For example, the processing circuit 120 estimates a disease of the medical treatment subject. The processing circuit 120 estimates information on medical treatment of the medical treatment subject by inputting the examination data and the reply data of the medical treatment subject to the second model MDL2 trained in advance. For example, the processing circuit 120 estimates a disease of the medical treatment subject. Then, the processing circuit 120 outputs the first estimation result representing the information on medical treatment estimated using the first model MDL1 (for example, a disease of the medical treatment subject) and the second estimation result representing the information on medical treatment estimated using the second model MDL2 (for example, a disease of the medical treatment subject) via the output interface 113. Accordingly, a medical staff member can treat patients without being affected by the instability of their replies in medical examinations by interview.
Although several embodiments have been described, these embodiments are presented as examples and are not intended to limit the scope of the invention. These embodiments can be implemented in various other embodiments, and various omissions, replacements, and changes can be made without departing from the gist of the invention. These embodiments and modifications thereof are included in the scope of the invention described in the claims and the equivalent scope thereof, as are included in the scope and gist of the invention.
Regarding the above embodiments, the following appendices are disclosed as one aspect and selective features of the invention.
A medical information processing system including a processing circuit configured:
The output unit may include a display unit. The processing circuit may display the first estimation result and the second estimation result side by side on the display unit.
The processing circuit may further compare the first estimation result with the second estimation result to determine whether or not the first estimation result and the second estimation result match.
The processing circuit may calculate a similarity between the first estimation result and the second estimation result. Further, the processing circuit may determine that the first estimation result and the second estimation result match if the similarity is equal to or greater than a threshold value and may determine that the first estimation result and the second estimation result do not match if the similarity is less than the threshold value.
The processing circuit may output an alert via the output unit if it is determined that the first estimation result and the second estimation result do not match.
The processing circuit may weight the reply data of the medical treatment subject at the time of inputting the reply data of the medical treatment subject to the second model. Further, the processing circuit may estimate information on medical treatment of the medical treatment subject by inputting the weighted reply data and the examination data of the medical treatment subject to the second model.
The processing circuit may repeatedly estimate information on medical treatment while changing a weighting factor. The processing circuit may compare the second estimation results representing a plurality of pieces of information on medical treatment repeatedly estimated using the second model to determine whether or not the plurality of second estimation results match.
The processing circuit may output the second estimation result for each weighting factor via the output unit.
The information on medical treatment may include estimating a disease of a patient.
A medical information processing method using a computer, including:
A non-transitory computer-readable storage medium storing a program for causing a computer to execute:
| Number | Date | Country | Kind |
|---|---|---|---|
| 2021-131303 | Aug 2021 | JP | national |
| 2022-122133 | Jul 2022 | JP | national |
The present application claims priority based on Japanese Patent Applications No. 2021-131303 filed Aug. 11, 2021 and No. 2022-122133 filed Jul. 29, 2022, the contents of which is incorporated herein by reference.
