Estimation device and display device

Abstract

To estimate a disease related to a sound acquired from a waveform of the sound caused by a target part in a living body.SOLUTION: The estimation device includes an acquisition unit, an input unit, and an approximator. The acquisition unit acquires information on a biological sound of a living body. Input information related to a biological sound of a living body is input to the input unit. The approximator can estimate an estimation result of a disease related to a sound caused by a target site from input information input to the input unit. The approximator has a trained parameter for obtaining an estimation result of a disease related to a sound caused by a target site from input information.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to an estimation device and a display device. [Background technology]

[0002] In a conventional device for estimating a disease from a body sound, for example, a technology is disclosed in which the body sound acquired from a subject is compared with standard data of body sounds of various diseases to diagnose the name of the subject's disease (see, for example, Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2002-165789 Summary of the Invention [Problem to be solved by the invention]

[0004] Such an estimation device is required to estimate with higher accuracy whether or not a subject has an abnormality. [Means for solving the problem]

[0005] An estimation device according to one embodiment of the present disclosure includes an acquisition unit that acquires a first waveform including a waveform of a sound caused by a target part of a living body, and an estimation unit that estimates a disease related to the sound caused by the target part from input information including at least the first waveform, wherein the estimation unit has learned parameters for obtaining an estimation result of the disease from the input information, and the learned parameters are set based on training data including a second waveform including the waveform of the sound caused by the target part and training data including information related to the disease of the living body having the second waveform.

[0006] A display device according to one embodiment of the present disclosure displays a first waveform including a waveform of a sound caused by a target part of a living body acquired by an acquisition unit, and a disease related to the sound caused by the target part estimated by an estimation unit from input information including at least the first waveform, wherein the estimation unit has learned parameters for obtaining an estimation result of the disease from the input information, and the learned parameters are set based on learning data including a second waveform including the waveform of the sound caused by the target part and teacher data including a disease of the living body having the second waveform. [Effects of the Invention]

[0007] According to an estimation device according to an embodiment of the present disclosure, it is possible to estimate with higher accuracy whether or not a subject has an abnormality. [Brief explanation of the drawings]

[0008] [Figure 1] 1 is a block diagram showing an example of a schematic configuration of a waveform acquisition device according to an embodiment of the present invention. [Figure 2] FIG. 2 is a diagram for explaining the operation of the estimation device according to the present embodiment. [Figure 3] FIG. 2 is a diagram illustrating an example of a configuration of a neural network in the estimation device according to the present embodiment. [Figure 4] 10 is a diagram showing an example of how acquired learning data and disease data are associated with each other in the estimation device according to the present embodiment. FIG. [Figure 5] FIG. 2 is a diagram for explaining the operation of the estimation device according to the present embodiment. [Figure 6] FIG. 1 is a diagram illustrating an example of an estimation system according to an embodiment of the present invention. [Figure 7] 10A and 10B are diagrams illustrating an example of an image displayed on a display unit in the estimation device according to the present embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0009] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings, but the present disclosure is not limited thereto. Also, for convenience of illustration, the dimensional ratios of the drawings do not necessarily correspond to those in the description. In the components shown in the following drawings, the same components are denoted by the same reference numerals.

[0010] 1 is a block diagram showing an example of the configuration of an estimation device 1 according to an embodiment of the present disclosure. The estimation device 1 functions as an estimation device that estimates a disease related to body sounds caused by a target part. The target part may be, for example, at least one of the respiratory system including the lungs, the circulatory system including the heart, the digestive system including the intestines, and the reproductive organs.

[0011] 1, the estimation device 1 includes components such as an acquisition unit 10, a control unit 20, a storage unit 30, a processing unit 40, a communication unit 50, a display unit 60, and an input unit 70. These components are electrically connected to each other, for example, by a bus 80. However, without being limited thereto, these components may be connected to each other via, for example, a local area network, the Internet network, etc., to transmit and receive information.

[0012] In this embodiment, the target living body is assumed to be a human, but the target living body may be any living body, including animals such as dogs, cats, and horses. In this embodiment, the user may include any person who uses information about body sounds, such as a medical professional, a medical student, or a related person who does not have medical skills but is involved in the acquisition of the information. Medical professionals may include, for example, doctors, dentists, pharmacists, public health nurses, midwives, nurses, licensed practical nurses, dental hygienists, radiological technologists, dental technicians, clinical laboratory technicians, medical laboratory technicians, physical therapists, occupational therapists, orthoptists, clinical engineers, prosthetists, emergency medical technicians, speech-language-hearing therapists, and registered dietitians.

[0013] 1, the lines connecting the functional blocks indicate the flow of control signals or information to be communicated. The communication between the functional blocks may be wired or wireless.

[0014] The estimation device 1 according to the embodiment of the present disclosure outputs disease estimation information related to the body sounds of a living body from target information, which is information related to the living body. The target information includes at least one of information related to the body sounds of the living body and health-related information.

[0015] The acquisition unit 10 may acquire target information. The acquisition unit 10 may acquire information related to body sounds of a living body. The acquisition unit 10 acquires a first waveform, which is acquired data including a waveform of a sound originating from a target part. The target part may be, for example, at least one of the respiratory system including the lungs, the circulatory system including the heart, the digestive system including the intestines, and the reproductive organs. The acquisition unit 10 may be equipped with a sensor that can acquire vibrations generated by body sounds by contacting the living body, and may use, for example, an angular velocity sensor, an acceleration sensor, a gyro sensor, or a strain sensor. The sensor may detect, for example, the voltage of vibrations generated by body sounds. The sensor may detect, for example, the frequency of vibrations generated by body sounds. The acquisition unit 10 includes one sensor, but is not limited to this. The acquisition unit 10 may include, for example, multiple sensors. The acquisition unit 10 may include, for example, multiple sensors capable of detecting different frequencies, or multiple sensors capable of detecting the same frequency. The acquisition unit 10 may include, for example, multiple sensors of different types. The acquiring unit 10 is not limited to a configuration for acquiring vibrations, and may, for example, collect body sounds using a microphone or the like and convert the acquired sounds into electronic data. In addition, the acquiring unit 10 may input information about body sounds acquired by a device other than the estimation device 1. For example, the acquiring unit 10 may read an image on which information about body sounds is written. The first waveform may be, for example, a waveform measured by an electrocardiogram or the like.

[0016] The acquisition unit 10 may be brought into contact with, for example, the chest, back, neck, abdomen, etc. of a living body. The acquisition unit 10 may be configured to receive vibrations generated by biological sounds through a membrane and detect the vibrations generated in the membrane. For example, by bringing the acquisition unit 10 into contact with the chest, the acquisition unit 10 can detect heart sounds and / or The acquisition unit 10 may detect vibrations due to bowel sounds and / or fetal heart sounds by, for example, contacting the back. The acquisition unit 10 may detect vibrations due to breathing sounds by, for example, contacting the neck. The acquisition unit 10 may detect vibrations due to bowel sounds and / or fetal heart sounds by, for example, contacting the abdomen. The acquisition unit 10 transmits a detection signal corresponding to the detected vibrations. The membrane receives vibrations generated by biological sounds. The membrane deforms in response to the received vibrations. The acquisition unit 10 may be an electronic stethoscope that acquires biological sounds. The electronic stethoscope may be, for example, a stethoscope that contacts a living body to acquire biological sounds, converts the acquired biological sounds into digital data, and outputs the digital data. The acquisition unit 10 may transmit the acquired first waveform to the processing unit 40. The acquisition unit 10 may convert the first waveform into sound and transmit the first sound to the processing unit 40.

[0017] The control unit 20 may generally manage the operation of the estimation device 1 by controlling the other components of the estimation device 1. The control unit 20 includes at least one processor to provide control and processing power for performing various functions.

[0018] According to this embodiment, the at least one processor may be implemented as a single integrated circuit (IC) or as multiple communicatively coupled integrated circuits (ICs) and / or discrete circuits. The processors may be implemented according to various known techniques.

[0019] In this embodiment, a processor includes one or more circuits or units configured to perform one or more data computational procedures or processes, for example, by executing instructions stored in associated memory. In this embodiment, a processor may also be firmware (e.g., discrete logic components) configured to perform one or more data computational procedures or processes.

[0020] According to this embodiment, the processor may be, for example, one or more processors, controllers, microprocessors, microcontrollers, application specific integrated circuits (ASICs), digital signal processors, programmable logic devices, or the like. The control unit 20 may include a network device, a field programmable gate array, or any combination of these devices or configurations, or a combination of other known devices and configurations, to perform the functions described below. In this embodiment, the control unit 20 includes, for example, a CPU (Central Processing Unit).

[0021] The storage unit 30 includes a non-transitory recording medium that can be read by the CPU of the control unit 20, such as a read-only memory (ROM) and a random access memory (RAM).

[0022] The storage unit 30 may store a control program 31 for controlling the estimation device 1. Various functions of the control unit 20 may be realized by the CPU of the control unit 20 executing the control program 31 in the storage unit 30. The control program 31 may be a disease estimation program related to body sounds. In this example, the control unit 20 executes the control program 31 in the storage unit 30, whereby an approximator 270 capable of outputting an estimation result 300 is configured in the control unit 20, as shown in FIG. 2. The approximator 270 includes, for example, a neural network 200. The control program 31 can also be said to be a program for causing the estimation device 1 to function as the neural network 200. An example configuration of the neural network 200 will be described later.

[0023] In addition to the control program 31, the storage unit 30 also stores trained parameters 32, training data 34, and teacher data 35 related to the neural network 200. The training data 34 and teacher data 35 are data used when training the neural network 200. The trained parameters 32 and estimation data 33 are data used when the trained neural network 200 estimates a disease related to body sounds. The storage unit 30 may also store estimation data 33 (hereinafter also referred to as "input information").

[0024] The training data 34 is data that is input to the input layer 210 of the neural network 200 when the neural network 200 is trained. The training data 34 is also called training data. The teacher data 35 is data that indicates the correct result of a disease related to body sounds. The training data 34 may include waveforms that are not suspected of being disease. The training data 34 may include waveforms that are suspected of being disease. The teacher data 35 is compared with the output data 400 output from the output layer 230 of the neural network 200 when the neural network 200 is trained.

[0025] The estimation data 33 is data input to the input layer 210 of the trained neural network 200 when the trained neural network 200 estimates a disease related to body sounds. The trained parameters 32 are trained parameters in the neural network 200. The trained parameters 32 can be said to be parameters adjusted by training the neural network 200. The trained parameters 32 include weighting coefficients that indicate the weights of connections between artificial neurons. As shown in FIG. 2 , the trained neural network 200 performs calculations based on the trained parameters 32 on the estimation data 33 input to the input layer 210, and outputs an estimation result 300 from the output layer 230.

[0026] The data input to the input layer 210 may be input to the input layer 210 via the input unit 70, or may be input directly to the input layer 210. When the data is input directly to the input layer 210, the input layer 210 may be a part or the whole of the input unit 70. Hereinafter, the disease estimation result related to body sounds may be referred to as the estimation result 300.

[0027] The memory unit 30 may store a second waveform. The second waveform may be data including a waveform of a sound originating from the target area. The second waveform may be a reference waveform for the first waveform (hereinafter, sometimes referred to as a reference waveform). The second waveform may be, for example, a normal waveform, which is an ideal waveform, or an abnormal waveform, which is a waveform different from the ideal waveform. The second waveform may be any of a waveform acquired by the acquisition unit 10, a waveform acquired by a device other than the estimation device 1, and a waveform created by a computer, etc. The normal waveform may be a waveform of the same person as the living body that measured the first waveform, or a waveform of a different person than the living body that measured the first waveform. The normal waveform may be a waveform that is not suspected of a disease. The normal waveform may be a waveform when the target area is in a normal state. The abnormal waveform may be a waveform that is suspected of a disease.

[0028] The second waveform may be stored in an external database. The storage unit 30 may read out the database to acquire the second waveform. The storage unit 30 may store at least one of the second waveform, the second sound, the type of the second waveform, and characteristic information of the second waveform in association with each other. The second sound may be sound data obtained by converting the second waveform into sound. The type of the second waveform may be, for example, a type of waveform such as respiratory system sounds including respiratory sounds, circulatory system sounds including heart sounds, digestive system sounds including bowel sounds, and fetal heart sounds.

[0029] The processing unit 40 may receive the first waveform from the acquisition unit 10. The processing unit 40 may receive information on the type of the first waveform from the input unit 70. The processing unit 40 may receive the estimation result from the storage unit 30. 300. The processing unit 40 may generate an image 61 to be transmitted to the display unit 60 based on the estimation result 300. The processing unit 40 may change the content of the generated image 61 depending on the user who acquires the information displayed on the display unit 60.

[0030] The processing unit 40 may search for the inference result 300 and similar reference data, and output the search results together with the inference result 300. The processing unit 40 may generate an image 61 to be sent to the display unit 60 based on the inference result 300 and the search results. For example, if similar reference data exists, the processing unit 40 may output the reference data as a search result. For example, if similar reference data does not exist, the processing unit 40 may output a message indicating that there is no reference data as a search result. The reference data may include learning data 34. The reference data may include teacher data 35. When the user obtaining the information displayed on the display unit 60 is a medical professional, the processing unit 40 may generate the image 61 based on the inference result 300 and the search results, for example.

[0031] When a predetermined condition is met while an application for acquiring the first waveform is running, the processing unit 40 may manually or automatically start and / or stop the acquisition of the first waveform. The predetermined condition for starting and / or stopping the acquisition of the first waveform will be described. The processing unit 40 may start the acquisition of the first waveform when a first condition is met. The first condition is a condition for starting the acquisition of the first waveform. Starting the acquisition of the first waveform may include, for example, temporarily suspending the acquisition of the first waveform and then resuming it.

[0032] The processing unit 40 may start acquiring the first waveform when, for example, a first condition is satisfied. The first condition may be, for example, that an application for acquiring the first waveform is started and vibration is detected by a sensor provided in the acquisition unit 10. The first condition may be, for example, that the application for acquiring the first waveform is started. At that time, the processing unit 40 may determine that the acquisition unit 10 is in contact with a living body.

[0033] The first condition may be, for example, that a predetermined time (for example, several seconds, specifically 2 to 5 seconds) has elapsed since vibration was detected by a sensor provided in the acquisition unit 10. For example, immediately after vibration begins to be detected by the sensor provided in the acquisition unit 10, the first waveform may not be stable due to, for example, the user adjusting the position at which the acquisition unit 10 is in contact. Therefore, by having the processing unit 40 start acquiring the first waveform when a predetermined time has elapsed since vibration detection began, it becomes easier to acquire a stable first waveform.

[0034] The first condition may be, for example, that the state in which vibration is detected by the sensor provided in the acquisition unit 10 continues for a predetermined time (for example, several seconds, specifically 2 to 5 seconds). For example, immediately after vibration is detected, the first waveform may not be stable because the user adjusts the position at which the acquisition unit 10 is in contact with the living body. If the user fixes the position at which the acquisition unit 10 is in contact with the living body, the acquired first waveform becomes more stable. The processing unit 40 starts acquiring the first waveform when the state in which vibration is detected continues for a predetermined time, thereby making it easier to acquire a stable first waveform. The first condition is not limited to the above example and may include other conditions.

[0035] The processing unit 40 may terminate the acquisition of the first waveform when, for example, the second condition is satisfied. The second condition may be, for example, that it is determined based on a sensor provided in the acquisition unit 10 that it is not possible to acquire the first waveform. The second condition may be, for example, that it is determined based on a sensor provided in the acquisition unit 10 that it is not possible to acquire the first waveform for a predetermined time or longer. The second condition may be, for example, that the processing unit 40 determines that the acquisition unit 10 is not in contact with a living body. The second condition may be, for example, that a predetermined time (for example, several seconds, specifically 2 to 5 seconds) has elapsed since the acquisition unit 10 started to acquire the first waveform. The second condition may be, for example, that it is determined based on a sensor provided in the acquisition unit 10 that it is not possible to acquire the first waveform for a predetermined time or longer. The second condition may be that the state in which the acquisition has started continues for a predetermined time (for example, several seconds, specifically 2 to 5 seconds). The second condition is not limited to the above example, and may include other conditions.

[0036] The processing unit 40 may receive information regarding the type of the first waveform input from the input unit 70. The processing unit 40 may determine the type of the first waveform. The processing unit 40 may analyze the characteristics of the first waveform (hereinafter referred to as characteristic information of the first waveform). The processing unit 40 may determine the type of the first waveform from the characteristic information of the first waveform.

[0037] Because the shape of the waveform varies depending on the type of biological sound, the processing unit 40 may, for example, determine the type of waveform based on the shape of the waveform. The processing unit 40 may, for example, determine the type of waveform due to biological sounds such as respiratory system sounds including breath sounds, circulatory system sounds including heart sounds, digestive system sounds including bowel sounds, and fetal heart sounds. The type of waveform may, for example, be selected by a user's operation. If the processing unit 40 cannot determine the type of waveform, the display unit 60 may, for example, display information prompting the user to acquire the first waveform again. If the processing unit 40 cannot determine the type of waveform, the display unit 60 may, for example, display information prompting the user to select the type by a user's operation.

[0038] In the following description, a case where the body sound is a heart sound will be described as an example. However, the estimation device 1 of the present invention is not limited to the above example and may be used for other body sounds. Other body sounds may be, for example, respiratory system sounds, circulatory system sounds, digestive system sounds, fetal heart sounds, etc.

[0039] For example, even if the processing unit 40 determines that the first waveform includes a waveform of a heart sound, the positional relationship between the acquisition unit 10 and the heart, which is the target region, varies depending on the region where the acquisition unit 10 is in contact with the living body. The shape of the acquired waveform changes with the change in the positional relationship between the acquisition unit 10 and the target region. Therefore, the region where the acquisition unit 10 is in contact may be determined from the shape of the first waveform. The region where the acquisition unit 10 is in contact may be, for example, the aortic valve region, which is the right edge of the second intercostal space at the sternal border; the pulmonary valve region, which is the left edge of the second intercostal space at the sternal border; the Erb region, which is the left edge of the third intercostal space at the sternal border; the tricuspid valve region, which is the left edge of the fourth intercostal space at the sternal border; and the mitral valve region, which is the apex of the heart. The region where the acquisition unit 10 is in contact may be selected by, for example, a user's operation.

[0040] The processing unit 40 may analyze the characteristic information of the first waveform. The processing unit 40 may receive the second waveform from the storage unit 30. The processing unit 40 may analyze the characteristic information of the second waveform. The characteristic information of the first waveform or / and the characteristic information of the second waveform may be information about the period of the first waveform or / and the second waveform. The characteristic information of the first waveform or / and the characteristic information of the second waveform may be information about the constant period of the first waveform or / and the second waveform. The characteristic information of the first waveform or / and the characteristic information of the second waveform The information may be information about the number of cycles per time. The processing unit 40 may, for example, perform a Fourier transform on the voltage of the vibration generated by the body sound acquired by the acquiring unit 10 to obtain the frequency.

[0041] A cardiac cycle is made up of diastole, when the heart expands, and systole, when the heart contracts, with one heartbeat. During systole, the first sound is generated, which is the sound of the atrioventricular valves closing. During diastole, the second sound is generated, which is the sound of the aortic valve and pulmonary valve closing. When the heart works normally, one heartbeat is one cycle, and it moves at a regular interval. When the heart does not work normally, it does not move at a regular interval and / or a heart murmur occurs. A heart murmur is a heart sound that does not occur in a normal heart. A period may be from the beginning of the first note to the end of the second note.

[0042] For example, in the case of heart sounds, a normal waveform may be a waveform in which the first and second sounds alternate at a regular interval. In this case, the waveform may be one in which there is a low possibility that a waveform suspected of being a heart murmur is included between the first and second sounds. An abnormal waveform may be, for example, a waveform in which the first and second sounds are not suspected of alternating at a regular interval and / or a waveform in which there is a waveform suspected of being a heart murmur between the first and second sounds. The abnormal waveform may be, for example, a waveform characteristic of a heart disease. The cardiac diseases may include myocardial infarction, myocarditis, cardiomyopathy, valvular heart disease, arrhythmia, etc. The valvular heart disease may include, for example, aortic valve stenosis, aortic valve regurgitation, mitral valve stenosis, mitral valve regurgitation, mitral valve prolapse, etc. The abnormal waveform may include, for example, a characteristic waveform due to heart failure.

[0043] For example, in the case of respiratory sounds, the normal waveform may be a waveform due to tracheal respiratory sounds, bronchial respiratory sounds, alveolar respiratory sounds, etc. The abnormal waveform may be a waveform due to abnormal sounds characteristic of respiratory diseases, such as snoring, whooshing, crepitus, and water bubbles.

[0044] The processing unit 40 may analyze, for example, the period of the waveform, the shape of the waveform, the number of periods per time, etc. The processing unit 40 may identify multiple periods and analyze periods that have different waveform characteristics compared to waveforms of other periods.

[0045] The processing unit 40 may cause the display unit 60 to display an image 61 created from the information acquired by the acquisition unit 10, the storage unit 30, and the input unit 70. The processing unit 40 may select the type of the second waveform based on the information related to the first waveform.

[0046] The processing unit 40 may cause the display unit 60 to display an image 61 showing a first waveform. The processing unit 40 may cause the display unit 60 to display an image 61 showing a second waveform. The processing unit 40 may cause the display unit 60 to display an image 61 showing at least two or more waveforms, such as a normal waveform and an abnormal waveform. The processing unit 40 may cause the display unit 60 to display an image 61 showing the first waveform and the second waveform. The processing unit 40 may cause the display unit 60 to display an image 61 showing only one cycle of the first waveform and / or the second waveform. The processing unit 40 may cause the display unit 60 to display an image 61 showing multiple cycles of the first waveform and / or the second waveform. The processing unit 40 may cause the display unit 60 to display an image 61 showing multiple cycles of the first waveform and / or the second waveform.

[0047] The first waveform and / or the second waveform may be displayed, for example, as a time waveform. In a time waveform, the horizontal axis may represent time and the vertical axis may represent voltage. The first waveform and / or the second waveform may be displayed, for example, as a spectrogram. In a spectrogram, the horizontal axis may represent time and the vertical axis may represent frequency. The spectrogram may show the intensity at a certain frequency at a certain point in time using brightness or color.

[0048] The processing unit 40 may cause the display unit 60 to display an image 61 that displays the first waveform and the second waveform superimposed on each other. The processing unit 40 may cause the display unit 60 to display an image 61 that displays the first waveform and the second waveform in different colors. The processing unit 40 may cause the display unit 60 to display an image 61 that displays the first waveform and the second waveform with different lines. The processing unit 40 determines whether the first waveform satisfies a third condition. If the third condition is not satisfied, the processing unit 40 may cause the display unit 60 to display an image 61 that displays at least a portion of the first waveform differently from the normal display. The different display refers to a display that is different from at least a portion of the first waveform that satisfies the third condition. The different display may involve superimposing an icon image 67 on a portion that does not satisfy the third condition. The third condition may be, for example, that the difference in voltage between the first waveform and the second waveform is within a predetermined range. The third condition may be, for example, that the difference in frequency between the first waveform and the second waveform is within a predetermined range. The third condition may be, for example, that the difference in frequency between the first waveform and the second waveform is within 1 kHz.

[0049] The processing unit 40 may cause the display unit 60 to display an image 61 that displays the first waveform and the second waveform side by side. The processing unit 40 may cause the display unit 60 to display an image 61 that displays the first waveform and the second waveform superimposed on each other. The processing unit 40 may cause the display unit 60 to display an image 61 that displays the first waveform and the second waveform superimposed on each other based on a fourth condition.

[0050] The fourth condition may be to align the time widths of the period (ΔT1) of the first waveform and the period (ΔT2) of the second waveform. The fourth condition may be to change the time width of the period (ΔT2) of the second waveform based on the period (ΔT1) of the first waveform. The fourth condition may be to change the time width of the period (ΔT1) of the first waveform based on the period (ΔT2) of the second waveform.

[0051] The processing unit 40 may, for example, measure the time of the period (ΔT1) of the first waveform and the period (ΔT2) of the second waveform, and align the time widths by multiplying the time data of the second waveform by (ΔT1 / ΔT2).

[0052] The fourth condition may be that the position where the first waveform reaches its maximum value and the position where the second waveform reaches its maximum value are aligned. The fourth condition may be that the position where the second waveform reaches its maximum value are changed based on the position where the first waveform reaches its maximum value. The fourth condition may be that the position where the first waveform reaches its maximum value are changed based on the position where the second waveform reaches its maximum value. For example, in the case of heart sounds, the fourth condition may be that the position where the first sound of the first waveform reaches its maximum value and the position where the first sound of the second waveform reaches its maximum value are aligned. For example, in the case of heart sounds, the fourth condition may be that the position where the second sound of the first waveform reaches its maximum value and the position where the second sound of the second waveform reaches its maximum value are aligned.

[0053] The fourth condition may be that the time widths of the cycles of the first waveform and the second waveform are aligned. In this case, the maximum value of the first waveform and the maximum value of the second waveform may be aligned. The fourth condition may be that the position at which the first waveform reaches its maximum value is aligned with the position at which the second waveform reaches its maximum value. In this case, the maximum value of the first waveform and the maximum value of the second waveform may be aligned.

[0054] When the adjustment is made based on the fourth condition, the processing unit 40 may cause the display unit 60 to display an image 61 indicating that the adjustment has been made. The processing unit 40 may cause the display unit 60 to display an image 61 indicating the type of the first waveform and / or second waveform.

[0055] The processing unit 40 may cause the display unit 60 to display an image 61 indicating a play button 63a that outputs a first sound. The processing unit 40 may cause the display unit 60 to display an image 61 indicating a play button 63b that outputs a second sound. When the play button 63a is selected, the processing unit 40 may output the first sound. When the play button 63b is selected, the processing unit 40 may output the second sound.

[0056] When the discriminator 21 is unable to discriminate the type of waveform, the processor 40 may cause the display unit 60 to display an image 61 prompting the user to acquire the first waveform again. The processor 40 may cause the display unit 60 to display an image 61 showing the estimation result 300. The processor 40 may cause the display unit 60 to display an image 61 prompting the user to input the first information.

[0057] The communication unit 50 is connected to a communication network 700, including the Internet, via a wired or wireless connection. The communication unit 50 is capable of communicating with other devices, such as a cloud server and a web server, through the communication network 700. The communication unit 50 is capable of inputting information received from the communication network 700 to the control unit 20. The communication unit 50 is also capable of outputting information received from the control unit 20 to the communication network 700.

[0058] The display unit 60 receives and displays the image 61 generated by the processing unit 40. The display unit 60 may be, for example, a terminal having a display function such as a display. The display unit 60 may be, for example, a terminal having a speaker that outputs sound in addition to a display function such as a display. The display unit 60 may be, for example, a terminal having a speaker that outputs sound and a microphone that inputs sound in addition to a display function such as a display. The terminal may be, for example, a personal computer, a tablet terminal, a smartphone, etc. The display unit 60 may include, for example, a liquid crystal display, an organic EL (Electroluminescence) display, etc. The display unit 60 may display an image 61 input from the processing unit 40 via a network, for example. The display unit 60 is controlled by the control unit 20 to display various types of information such as characters, symbols, and figures.

[0059] The input unit 70 may receive target information as input. The input unit 70 may receive health-related information as input. The health-related information may include, for example, at least one of heart rate information, pulse rate information, pulse wave information, blood glucose level information, blood pressure information, blood oxygen saturation information, blood flow rate information, body temperature information, skin temperature information, sleep time information, age information, gender information, height information, weight information, drinking habit information, smoking habit information, and disease presence / absence information. The input unit 70 is capable of receiving input from a user to the estimation device 1. For example, information regarding body sounds measured by an external measurement device may be input to the input unit 70. The input unit 70 includes, for example, a keyboard and a mouse. The input unit 70 may include a touch panel capable of detecting user operations on the display surface of the display unit 60.

[0060] The configuration of the estimation device 1 is not limited to the above example. For example, the control unit 20 may include multiple CPUs. The control unit 20 may also include at least one DSP. All or some of the functions of the control unit 20 may be realized by a hardware circuit that does not require software to realize the function. The storage unit 30 may also include a computer-readable non-transitory recording medium other than ROM and RAM. The storage unit 30 may also include, for example, a small hard disk drive and an SSD (Solid State Drive). The storage unit 30 may also include a memory such as a USB (Universal Serial Bus) memory that is detachable from the estimation device 1. Hereinafter, the memory attached to the estimation device 1 will be referred to as a Removable memory is sometimes called "removable memory."

[0061] <Neural network configuration example> FIG. 3 is a diagram illustrating an example of the configuration of a neural network 200. In this example, the neural network 200 is, for example, a convolutional neural network (CNN). As shown in FIG. 3, the neural network 200 includes, for example, an input layer 210, a hidden layer 220, and an output layer 230. The hidden layer 220 is also called an intermediate layer. The hidden layer 220 includes, for example, a plurality of convolutional layers 240, a plurality of pooling layers 250, and a fully connected layer 260. In the neural network 200, the fully connected layer 260 is located before the output layer 230. In the neural network 200, the convolutional layers 240 and the pooling layers 250 are alternately arranged between the input layer 210 and the fully connected layer 260.

[0062] The configuration of neural network 200 is not limited to the example shown in Fig. 3. For example, neural network 200 may include one convolutional layer 240 and one pooling layer 250 between input layer 210 and fully connected layer 260. Furthermore, neural network 200 may be a neural network other than a convolutional neural network.

[0063] <Examples of estimation data, learning data, and teacher data> The estimation data 33 may include acquired data including a waveform of a sound originating from a target body part acquired by the acquisition unit 10. The estimation data 33 may include, for example, acquired data including electronic data of a sound acquired by the acquisition unit 10. The estimation data 33 may include, for example, acquired data related to body sounds acquired by a device other than the acquisition unit 10. The estimation data 33 may include, for example, image data in which information related to body sounds is described.

[0064] Hereinafter, the acquired data included in the estimation data 33 may be referred to as "acquired data for estimation." Also, the acquired data included in the estimation data 33 may be referred to as "acquired waveform for estimation." The acquired data included in the learning data 34 may also be referred to as "learning acquired data." The acquired data included in the learning data 34 may also be referred to as "learning acquired waveform." The learning data 34 includes a plurality of learning acquired data, each representing a plurality of learning acquired waveforms.

[0065] The estimation data 33 may be, for example, acquired data including a waveform of heart sounds obtained by contacting the acquisition unit 10 to the chest, acquired data including a waveform of respiratory sounds obtained by contacting the acquisition unit 10 to the chest, acquired data including a waveform of respiratory sounds obtained by contacting the acquisition unit 10 to the neck, acquired data including a waveform of bowel sounds obtained by contacting the acquisition unit 10 to the abdomen, and acquired data including a waveform of fetal heart sounds obtained by contacting the acquisition unit 10 to the abdomen. If the estimation acquired waveform includes a periodic waveform, for example, if it is determined that there is at least a waveform with a different periodicity, the waveform with at least a different periodicity may be input.

[0066] The training data 34 includes at least one type of acquired data including a waveform of heart sounds acquired by contacting the acquisition unit 10 to the chest, acquired data including a waveform of respiratory sounds acquired by contacting the acquisition unit 10 to the chest, acquired data including a waveform of respiratory sounds acquired by contacting the acquisition unit 10 to the neck, acquired data including a waveform of bowel sounds acquired by contacting the acquisition unit 10 to the abdomen, and acquired data including a waveform of fetal heart sounds acquired by contacting the acquisition unit 10 to the abdomen. The training data 34 may include some or all of the types of acquired data. When the acquired training waveform includes a periodic waveform, for example, when it is determined that there is a waveform with a different periodicity in part, at least the waveform with the different periodicity may be input.

[0067] The training data 35 includes, for each of the multiple training data 34 included in the training data 34, disease data related to the body sounds of a person having the training acquired waveform represented by the training data 34. The disease data related to the body sounds may be, for example, at least one of the presence or absence of a disease related to the body sounds, the name of the disease, and the severity of the disease. When the training data 34 are heart sounds, the training data 35 may be, for example, at least one of the presence or absence of a heart disease such as myocardial infarction, myocarditis, cardiomyopathy, valvular heart disease, and arrhythmia. Valvular heart disease includes, for example, at least one of aortic valve stenosis, aortic valve regurgitation, mitral valve stenosis, mitral valve regurgitation, and mitral valve prolapse. When the training data 34 are heart sounds, the training data 35 may be, for example, heart failure. When the training data 34 are respiratory sounds, the training data 35 may be, for example, at least one of the presence or absence of a respiratory disease causing snoring, whooshing, crepitus, blubbering, and the like. The disease name is not limited to one, and multiple names may be held. When the training data 34 is heart sounds, the teacher data 35 may be at least one of cardiac disease names such as myocardial infarction, myocarditis, cardiomyopathy, valvular heart disease, and arrhythmia. Valvular heart disease includes at least one of aortic valve stenosis, aortic valve regurgitation, mitral valve stenosis, mitral valve regurgitation, and mitral valve prolapse. When the training data 34 is heart sounds, the teacher data 35 may be, for example, heart failure. When the training data 34 is respiratory sounds, the teacher data 35 may be, for example, at least one of snores, whistles, crepitus, and bubbling sounds. When the training data 34 is respiratory sounds, the teacher data 35 may be, for example, at least one of respiratory disease names associated with snores, whistles, crepitus, and bubbling sounds.

[0068] The severity of a disease may be based on, for example, original standards or already known guidelines. In the case of the severity of a disease, the training data 35 may be, for example, the severity determined by a doctor's diagnosis, or a stage according to a predetermined guideline. The severity may be, for example, severe, moderate, or mild. Alternatively, the severity may be set to, for example, multiple stages. For example, in the case of heart failure, the severity may be set to stages A to D as follows based on at least one of the risk of heart failure, the structural condition of the heart (at least one structure of the myocardium, pericardium, and valve function), and the symptoms of heart failure.

[0069] Stage A: A condition in which there are high blood pressure, diabetes, coronary artery disease, etc., but no structural abnormalities in the myocardium, pericardium, or valve function, and there are no symptoms.

[0070] Stage B: Structural abnormalities such as left ventricular hypertrophy, cardiac enlargement, decreased cardiac function, valvular disease, and at least one history of myocardial infarction are present, but the patient is asymptomatic.

[0071] Stage C: Structural abnormalities are present and symptoms appear.

[0072] Stage D: Structural abnormalities are present and symptoms persist at rest despite adequate medical treatment.

[0073] The training data 34 includes acquired data from a plurality of different individuals. As shown in Fig. 4, in the storage unit 30, each of the plurality of acquired training data included in the training data 34 is associated with disease data of the individual having the acquired training data. Each of the plurality of acquired training waveforms used in training the neural network 200 may be associated with disease data of the individual having the body sound of the acquired training waveform. The disease data associated with the acquired training data is disease data diagnosed for the same individual as the individual having the body sound of the acquired training waveform, at approximately the same time as the time when the acquired training waveform indicated by the acquired training data was acquired.

[0074] Furthermore, the multiple acquired waveforms for learning, each indicated by the multiple acquired data for learning included in the learning data 34, may include waveform data of the same type as the acquired waveform for estimation, or may include waveform data of a different type from the acquired waveform for estimation.

[0075] If the control unit 20 determines that the waveform has been disturbed due to a change in the contact state between the living body and the acquisition unit 10 while acquiring the first waveform, the control unit 20 may exclude the waveform for the period in which it has been determined that the waveform has been disturbed. The control unit 20 may determine that the waveform has been disturbed, for example, when there is no change in the first waveform for a predetermined time (for example, several seconds, specifically 2 to 5 seconds). A change in the contact state may be, for example, a change in the positional relationship between the living body and the acquisition unit 10.

[0076] The control unit 20 of the estimation device 1 may acquire the acquired data for learning and the acquired data for estimation from the acquisition unit 10, or may acquire the data from a device other than the estimation device 1.

[0077] <Neural network learning example> FIG. 5 is a diagram illustrating an example of training of the neural network 200. When training the neural network 200, the control unit 20 inputs training data 34 to the input layer 210 of the neural network 200 as shown in FIG. 5. The control unit 20 then adjusts the variable parameters 32a in the neural network 200 so as to reduce the error of the output data 400 output from the output layer 230 of the neural network 200 relative to the training data 35. The control unit 20 inputs each acquired training data in the memory unit 30 to the input layer 210. When inputting the acquired training data to the input layer 210, the control unit 20 inputs multiple frequency data constituting the acquired training data to multiple artificial neurons constituting the input layer 210, respectively. The control unit 20 then adjusts the parameters 32a so as to reduce the error of the output data 400 output from the output layer 230 when the acquired training data is input to the input layer 210 relative to the result of the disease data corresponding to the acquired training data. As a method for adjusting the parameters 32a, for example, the backpropagation method is adopted. The adjusted parameters 32a become the trained parameters 32 and are stored in the storage unit 30. The parameters 32a include, for example, parameters used in the hidden layer 220. Specifically, the parameters 32a include the parameters used in the convolutional layer 220. The parameters 32a include filter coefficients used in the fully connected layer 40 and weighting coefficients used in the fully connected layer 260. However, the method for adjusting the parameters 32a, in other words, the method for learning the parameters 32a, is not limited to this.

[0078] The memory unit 30 stores learned parameters 32 that are obtained by learning the relationship between learning data 34, which includes image data of multiple acquired waveforms for learning, and disease data as teacher data 35, using a neural network 200.

[0079] In the above example, the estimation device 1 trains the neural network 200, but another device may train the neural network 200. In this case, the memory unit 30 of the estimation device 1 stores the trained parameters 32 generated by the other device. Furthermore, the memory unit 30 does not need to store the training data 34 and the teacher data 35. The communication unit 50 may receive the trained parameters 32 generated by the other device via the communication network 700, and the control unit 20 may store the trained parameters 32 received by the communication unit 50 in the memory unit 30. Alternatively, the trained parameters 32 generated by the other device may be stored in a removable memory included in the memory unit 30.

[0080] The neural network 200 trained as described above receives input of a plurality of acquired training waveform data as training data 34 into the input layer 210. Furthermore, the neural network 200 includes learned parameters 32 trained using disease data as training data 35. As shown in FIG. 2 , the neural network 200 performs calculations based on the learned parameters 32 on the estimation data 33 input to the input layer 210, and outputs an estimation result 300 from the output layer 230. When the acquired estimation data as the estimation data 33 is input to the input layer 210, a plurality of frequency data constituting the acquired estimation data are input to a plurality of artificial neurons constituting the input layer 210. The convolutional layer 240 performs calculations using filter coefficients included in the trained parameters 32, and the fully connected layer 260 performs calculations using weighting coefficients included in the trained parameters 32.

[0081] For example, when acquired estimation data indicating a heart sound is input to the input layer 210, an estimation result 300 of a person having the heart sound indicated by the acquired estimation data is output from the output layer 230. Furthermore, when acquired estimation data indicating a respiratory sound is input to the input layer 210, an estimation result 300 of a person having the respiratory sound indicated by the acquired estimation data is output from the output layer 230.

[0082] The storage unit 30 may store a plurality of pieces of estimation data 33. The control unit 20 inputs each of the plurality of pieces of estimation data 33 in the storage unit 30 to the input layer 210 of the neural network 200, and the output layer 230 of the neural network 200 outputs an estimation result 300 corresponding to each piece of estimation data 33.

[0083] As described above, in this example, acquired body sound data is used to train the neural network 200 and estimate a disease in the neural network 200.

[0084] The estimation result 300 obtained by the estimation device 1 may be displayed on the display unit 60. Furthermore, the estimation result 300 obtained by the estimation device 1 may be used in another device.

[0085] Fig. 6 is a diagram showing an example of a disease inference system 600 including an inference device 1 and a processing device 500 that performs processing using an inference result 300 obtained by the inference device 1. In the example of Fig. 6, the inference device 1 and the processing device 500 can communicate with each other via a communication network 700. The communication network 700 includes, for example, at least one of a wireless network and a wired network. The communication network 700 includes, for example, a wireless LAN (Local Area Network) and the Internet.

[0086] In the estimation device 1, a communication network 700 is connected to the communication unit 50. The control unit 20 causes the communication unit 50 to transmit the estimation result 300 to the processing device 500. The processing device 500 performs processing using the estimation result 300 received from the estimation device 1 via the communication network 700. For example, the processing device 500 is a display device such as a liquid crystal display device, and displays the estimation result 300. In this case, the processing device 500 may display the estimation result 300 in the form of a table or a graph. Furthermore, when multiple estimation devices 1 are connected to the communication network 700, the processing device 500 may display the estimation results 300 obtained by the multiple estimation devices 1. The configuration of the processing device 500 may be the same as or different from the configuration of the estimation device 1 shown in FIG. 1.

[0087] The processing performed by the processing device 500 using the estimation result 300 is not limited to the above example. The processing device 500 may communicate with the estimation device 1 directly via wireless or wired communication without using the communication network 700.

[0088] The estimation device 1 according to this embodiment obtains an estimation result of a disease related to body sounds from input information including at least a first waveform. The estimation device 1 can estimate the presence or absence of an abnormality in a subject by obtaining an estimation result of a disease related to body sounds from the input information. As a result, the estimation device 1 can grasp the risk of a disease related to body sounds, allowing a user (doctor) or the subject to intervene in early treatment.

[0089] <Other examples of estimation data and learning data> <Another example of the first> In this example, the training data 34 may include, for each training acquired data, target information including at least one piece of information relating to the health condition of a person whose bones are reflected in the training acquired waveform indicated by the training acquired data. The target information may also include acquisition time information corresponding to each piece of information included in the target information. The acquisition time information may be the time or point in time when each piece of information included in the target information was acquired, the date on which the body sound was acquired, the date on which the body was diagnosed, and the date on which each piece of information was entered into an electronic medical record, etc. The training data 34 includes, for each training acquired data, information relating to the health condition of the living body represented by the training acquired waveform indicated by the training acquired data. Hereinafter, information relating to the health condition of the living body represented by the training acquired waveform may be referred to as "training health-related information." Furthermore, information relating to the health condition of the living body represented by the training acquired waveform indicated by the training acquired data may be referred to as the training health-related information corresponding to the training acquired data.

[0090] The health-related information for learning includes at least one type of information, such as heart rate information, pulse rate information, pulse wave information, blood glucose level information, blood pressure information, blood oxygen saturation concentration information, blood flow rate information, body temperature information, skin temperature information, sleep time information, age information, sex information, height information, weight information, drinking habits information, smoking habits information, and disease presence / absence information. The health-related information for learning is compiled into a database for each person and generated as a CSV (Comma-Separated Value) file or a text file. Each of the heart rate information, pulse rate information, pulse wave information, blood glucose level information, blood pressure information, blood oxygen saturation concentration information, blood flow rate information, body temperature information, skin temperature information, sleep time information, age information, height information, and weight information is represented, for example, as multi-bit numerical data. In addition, in the gender information, for example, "male" or "female" is represented by 1-bit data, and in the drinking habit information, "drinking habit" or "not drinking habit" is represented by 1-bit data. In the smoking habit information, One bit of data is represented by "smoking" or "not smoking." Information on the presence or absence of a disease is represented by one bit of data by "suspected of having a disease" or "not suspected of having a disease." The health-related information for learning may also include the subject's body fat percentage or subcutaneous fat percentage.

[0091] When the training data 34 includes acquired training data and corresponding health-related information for training, disease data corresponding to the acquired training data (see FIG. 4 ) is also associated with the health-related information for training corresponding to the acquired training data. During training of the neural network 200, the acquired training data and the corresponding health-related information for training are simultaneously input to the input layer 210. Specifically, the acquired training data is input to some of the artificial neurons constituting the input layer 210, and the health-related information for training is input to other parts of the artificial neurons. Output data 400 output from the output layer 230 when the acquired training data and the corresponding health-related information for training are input to the input layer 210 is compared with the disease data corresponding to the acquired training data and the health-related information for training.

[0092] In this example, the estimation data 33 may include target information including at least one piece of acquired estimation data and information relating to the health condition of a person having an acquired estimation waveform indicated by the acquired estimation data. The target information may also include acquisition time information corresponding to each piece of information included in the target information. The acquisition time information may be the time or point in time when each piece of information included in the target information was acquired, the date on which the body sound was acquired, the date on which the body was diagnosed, and the date on which each piece of information was entered into an electronic medical record, etc. The estimation data 33 includes acquired estimation data and information relating to the health condition of the living body indicated by the acquired estimation waveform. Hereinafter, information relating to the health condition of the living body indicated by the acquired estimation waveform may be referred to as "health-related information for estimation." Furthermore, information relating to the health condition of the living body indicated by the acquired estimation data may be referred to as health-related information for estimation corresponding to the acquired estimation data.

[0093] Like the health-related information for learning, the health-related information for estimation includes at least one of the following types of information: heart rate information, pulse rate information, pulse wave information, blood glucose level information, blood pressure information, blood oxygen saturation information, blood flow rate information, body temperature information, skin temperature information, sleep time information, age information, gender information, height information, weight information, drinking habit information, smoking habit information, and disease presence / absence information. The health-related information for estimation includes the same types of information as the health-related information for learning. Similarly to the health-related information for learning, the health-related information for estimation may also include the subject's body fat percentage or subcutaneous fat percentage. The health-related information for estimation may be input from input unit 70.

[0094] In this example, when bone mineral density is estimated, acquired data for estimation and corresponding health-related information for estimation are simultaneously input to input layer 210. Specifically, acquired data for estimation is input to some of the multiple artificial neurons that make up input layer 210, and health-related information for estimation is input to other parts of the multiple artificial neurons. When acquired data for estimation and health-related information for estimation about a certain person are input to input layer 210, an estimation result 300 for that certain person is output from output layer 230.

[0095] The processing unit 40 may select the type of the second waveform based on the first waveform and the subject information. The processing unit 40 may, for example, change the criteria for determining whether a waveform is normal or abnormal depending on age information. The processing unit 40 may, for example, change the criteria for determining whether a waveform is normal or abnormal depending on gender information.

[0096] The image 61 generated by the processing unit 40 in the estimation device 1 of this embodiment will be described with reference to Fig. 7. The processing unit 40 may generate the image 61 that makes it easy to grasp the first waveform acquired by the acquisition unit 10, for example.

[0097] The waveform name 62 may indicate, for example, the type of waveform determined by the processing unit 40 .

[0098] The play button 63a outputs the first sound. The play button 63b outputs the second sound. When the play buttons 63a and 63b are selected, sound data is output. For example, the play buttons 63a and 63b may play a sound obtained by converting a waveform acquired by the acquisition unit 10. For example, the play buttons 63a and 63b may play a sound obtained by converting a waveform stored in the storage unit 30. For example, the play buttons 63a and 63b may play a sound acquired by the acquisition unit 10. For example, the play buttons 63a and 63b may play a sound stored in the storage unit 30. The processing unit 40 may display an image 61 displaying the play button 63a on the display unit 60. The processing unit 40 may display an image 61 displaying the play button 63b on the display unit 60. The processing unit 40 may display an image 61 displaying the play button 63a and / or the play button 63b on the display unit 60.

[0099] The displayed waveform 64 may display a first waveform. The displayed waveform 64 may display a second waveform. The displayed waveform 64 may display a first waveform and / or a second waveform. The displayed waveform 64 may, for example, display multiple second waveforms. The displayed waveform 64 may, for example, display at least two or more of a normal waveform and an abnormal waveform. The displayed waveform 64 may, for example, be displayed as a time waveform. The displayed waveform 64 may, for example, be displayed as a spectrogram. In FIG. 7, the first waveform and the second waveform are displayed overlapping each other, but the displayed waveform 64 may be displayed individually or may not be displayed depending on the waveform selection. The displayed waveform 64 may be displayed after the first waveform has been acquired for a predetermined time. In FIG. 7, the displayed waveform 64 displays only one cycle, but multiple cycles may be displayed. Among the multiple cycles, a cycle with waveform characteristics different from the other waveforms and a normal cycle may be displayed.

[0100] The estimation result comment 65 may display the estimation result 300. The estimation result comment 65 may be displayed based on, for example, the content of the estimation result 300. The estimation result comment 65 may display, for example, the presence or absence of a disease related to the acquired body sound, the name of at least one type of disease, the severity of the disease, etc.

[0101] The icon image 66 may be superimposed at a location where the first waveform differs from the second waveform by more than a third condition. The icon image 66 may be any shape that makes it easy to understand the difference between the first waveform and the second waveform. The icon image 67 may be any shape, such as a circle or a square.

[0102] Although the embodiments of the present disclosure have been described based on the drawings and examples, it should be noted that those skilled in the art would easily be able to make various modifications or alterations based on the present disclosure. Therefore, it should be noted that these modifications and alterations are included within the scope of the present disclosure. For example, the functions included in each component or step can be rearranged so as not to cause logical inconsistencies, and multiple components or steps can be combined or divided into one. The embodiments of the present disclosure can also be realized as a program executed by a processor included in an apparatus or a storage medium on which a program is recorded. It should be understood that these are also included within the scope of the present disclosure. [Explanation of symbols]

[0103] 1 Estimation device 10 Acquisition Department 20 Control Unit 30 Storage section 31 Control Program 32 trained parameters 32a Parameters 33 Estimation data 34 Training data 35 Training data 40 Processing section 50 Communications Department 60 Display section 61 images 62 Waveform name 63a, 63b Play button 64 Display waveform 65 Estimation result comments 66 Icon Images 70 Input section 80 Hub 200 Neural Networks 210 Input Layer 220 Hidden Layer 230 Output Layer 240 convolutional layers 250 Boolean Layer 260 fully connected layer 270 Approximator 300 Estimation results 400 output data 500 processing equipment 600 Disease Prediction System 700 Communication Network

Claims

1. an acquisition unit that acquires a first waveform including a waveform of a sound originating from a target part in a living body; an estimation unit that estimates a disease related to a sound originating from the target region from input information including at least the first waveform, the estimation unit has learned parameters for obtaining an estimation result of the disease from the input information, the learned parameters are set based on learning data including a second waveform that includes a waveform of a sound caused by the target region, and teacher data including information on the disease of a living body having the second waveform. Estimation device.

2. The teacher data includes a plurality of diseases of a living body having the second waveform. The estimation device according to claim 1 .

3. The training data includes the severity of the disease of the living body having the second waveform. The estimation device according to claim 1 .

4. the learned parameters are set based on learning data including information about the health state of a living organism having the first waveform. The estimation device according to claim 1 .

5. a discrimination unit that discriminates the type of the first waveform; the discriminator discriminates the type of waveform of any one of heart sounds, respiratory sounds, and bowel sounds; The estimation device according to claim 1 .

6. The estimation device according to claim 1 , wherein the training data includes at least one waveform of heart sounds, respiratory sounds, and bowel sounds.

7. the estimation unit uses learned parameters including a waveform of the same type as the first waveform identified by the identification unit. The estimation device according to claim 5 .

8. a display unit that displays the first waveform and the reference waveform; The reference waveform is a waveform for reference regarding the target region, and is at least one of a normal waveform and a waveform different from the normal waveform. The estimation device according to claim 1 .

9. the display unit displays the first waveform and the reference waveform in a superimposed manner. The estimation device according to claim 8 .

10. The first waveform and the reference waveform are displayed such that the position where the first waveform takes its maximum value and the position where the reference waveform takes its maximum value are aligned. The estimation device according to claim 8 .

11. the display unit displays the first waveform and the reference waveform differently. The estimation device according to claim 8 .

12. the display unit displays the first waveform and the reference waveform side by side. The estimation device according to claim 8 .

13. The display unit displays the estimation result estimated by the estimation unit. The estimation device according to claim 8 .

14. a processing unit that determines whether the first waveform and the reference waveform differ by more than a predetermined condition; the display unit displays at least a part of the first waveform determined by the processing unit to be different from the first waveform and the reference waveform that are not determined to be different. The estimation device according to claim 12.

15. a first waveform including a waveform of a sound originating from a target part in a living body acquired by an acquisition unit; displaying a disease related to a sound originating from the target region from input information including at least the first waveform estimated by the estimation unit; the estimation unit has learned parameters for obtaining an estimation result of the disease from the input information, the learned parameters are set based on learning data including a second waveform that includes a waveform of a sound caused by the target region, and teacher data including information on the disease of a living body having the second waveform. Display device.

Images