Medical information processing device, medical information processing method, and program

Abstract

To provide a medical information processing apparatus, a medical information processing method and a program that appropriately determine whether or not a measured value measured by a patient or the like is normal.SOLUTION: In a medical information processing apparatus 1, an acquisition function 13 acquires biological information associated with a patient. A detection function 15 detects tendency change of biological information at predetermined timing on the basis of time series data associated with the biological information. A determination function 18 determines whether the tendency change is due to a measurement mistake of the biological information or due to state change of the patient. A presentation apparatus 2 presents a determination result of the determination function 18.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The embodiments disclosed in the present specification and drawings relate to a medical information processing device, a medical information processing method, and a program. [Background technology]

[0002] To accurately and quickly grasp changes in physical condition, continuous home monitoring based on easily measurable patient information (e.g., blood pressure, transcutaneous arterial oxygen saturation, body temperature, pulse measured from facial images, etc.) is desirable. However, at home, measurements are performed by the patient themselves, who are not medical professionals, and various measurement abnormalities such as misalignment of the device and body movement are likely to occur. Monitoring using information that includes such measurement abnormalities can easily lead to misinterpretation of the measurement results, making it difficult to accurately grasp the patient's physical condition. In response to this, a health management system has been proposed that detects abnormal vital signs measured due to, for example, drinking alcohol or lack of sleep, and determines that the fluctuations are not due to disease. Also, a technology has been proposed that determines abnormalities in the measurement environment based on abnormal correlations between patient measurements. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2014-2687 [Patent Document 2] Japanese Patent Application Publication No. 2020-162649 Summary of the Invention [Problem to be solved by the invention]

[0004] The problem to be solved by the embodiments disclosed in this specification and the drawings is to appropriately determine whether or not a measurement value measured by a patient or the like is normal. However, the problem to be solved by the embodiments disclosed in this specification and the drawings is not limited to the above problem. Problems corresponding to the effects of each configuration shown in the embodiments described below can also be positioned as other problems. [Means for solving the problem]

[0005] The medical information processing apparatus of the present embodiment includes an acquisition unit, a detection unit, and a discrimination unit. The acquisition unit acquires biological information related to a patient. The detection unit detects a trend change in the biological information at a predetermined timing based on time-series data related to the biological information. The discrimination unit discriminates whether the trend change is due to a measurement error in the biological information or a change in the patient's condition. [Brief explanation of the drawings]

[0006] [Figure 1] FIG. 1 is a diagram showing an example of the arrangement of a medical image processing apparatus according to an embodiment. [Figure 2] FIG. 10 is a diagram showing an example of information stored in a patient DB. [Figure 3] FIG. 10 is a diagram for explaining an example of identifying similar patients from a patient DB. [Figure 4] FIG. 10 is a diagram for explaining an example of a method for calculating a propensity score performed by the first calculation function and the second calculation function. [Figure 5] FIG. 10 is a diagram showing examples of propensity scores calculated by the first calculation function and the second calculation function. [Figure 6] FIG. 10 is a diagram for explaining a method for calculating a propensity score when multiple conditions of similar patients are extracted from a patient DB. [Figure 7] FIG. 10 is a diagram for explaining how to calculate a propensity score using principal component analysis. [Figure 8] FIG. 10 is a diagram for explaining how to calculate a propensity score using maximum likelihood estimation. [Figure 9] FIG. 4 is a diagram for explaining a process performed by a determination function. [Figure 10] 10 is a flowchart showing an example of a procedure for calculating a propensity score and a discrimination process. [Figure 11] FIG. 10 is a diagram showing an example of a first propensity score and an example of a second propensity score for each case. [Figure 12] 10 is a flowchart of an example of a processing procedure performed by the medical image processing apparatus according to the embodiment. [Figure 13] FIG. 1 is a diagram for explaining an example of a usage scene of a medical information processing apparatus. [Figure 14] FIG. 10 is a diagram showing an example of information presented when it is determined that there are no measurement errors and no disease onset. [Figure 15] FIG. 10 is a diagram showing an example of information presented when it is determined that there is a measurement error but no disease has developed. [Figure 16] FIG. 10 is a diagram showing an example of information presented when it is determined that a disease has developed. [Figure 17] FIG. 10 is a diagram showing an example of the configuration of a medical image processing apparatus according to a modified example. [Figure 18] FIG. 10 is a diagram showing an example of an image presented on a presentation device. DETAILED DESCRIPTION OF THE INVENTION

[0007] Hereinafter, a medical image processing apparatus according to an embodiment will be described with reference to the drawings.

[0008] [Configuration of medical information processing device] Fig. 1 is a diagram showing an example of the configuration of a medical information processing device of this embodiment. As shown in Fig. 1, the medical information processing device 1 includes a patient DB (database) 11, a processing circuitry 12, an output interface 19, and a storage unit 20. The processing circuitry 12 includes an acquisition function 13, an identification function 14, a detection function 15, and a discrimination function 18. The detection function 15 includes a first calculation function 16 and a second calculation function 17.

[0009] The medical information processing device 1 refers to the information stored in the patient DB 11 and classifies the acquired biological information of the target patient into abnormal measurement, normal health, and disease onset based on the similarity of the tendency with the biological information of other patients, and outputs presentation information to the presentation device 2. The presentation device 2 is connected to the medical information processing device 1.

[0010] The presentation device 2 is, for example, at least one of an image display device, a printing device, a smartphone, a tablet terminal, etc. The presentation device 2 presents presentation information output by the medical information processing device 1.

[0011] The patient DB11 stores patient information for each of multiple patients other than the target patient (hereinafter simply referred to as "patient"). The patient information includes, for example, patient identification information, patient disease information, patient age information, patient gender information, and patient biometric information (measurement value information). The patient biometric information includes, for example, blood pressure, pulse rate, respiratory rate, and percutaneous arterial oxygen saturation value. The patient DB11 may store all or part of the patient information for each patient.

[0012] The acquisition function 13 acquires patient information of the patient to be discriminated. The patient information includes, for example, the patient's biometric information and the patient's attribute information. The patient's biometric information includes, for example, blood pressure, pulse rate, respiratory rate, percutaneous arterial oxygen saturation (SpO2) value, etc. The patient's attribute information includes, for example, age, gender, and disease history. The acquisition function 13 continuously acquires the patient's biometric information, for example, at each sampling time. The sampling time may be, for example, every second, every minute, every hour, every day, morning, noon, and night. There may be two or more types of biometric information. Note that, if attribute information is stored in the patient DB 11 as part of the patient information, the acquisition function 13 may acquire the attribute information from the patient DB 11. The acquisition function 13 outputs the acquired patient information of the patient to the identification function 14, the first calculation function 16, and the second calculation function 17.

[0013] The identification function 14 refers to the patient attribute information output by the acquisition function 13 and identifies similar patients from the patient DB 11 whose attributes are similar to those of the target patient, and extracts or calculates the measurement value distribution of the identified similar patients. Note that a similar patient is a patient in a group of patients with the same disease who has attributes closest to those of the target patient, for example, a patient of the same gender and the same age or within a specified range. Furthermore, when multiple similar patients are extracted, the identification function 14 extracts a measurement value distribution for each similar patient. The identification function 14 calculates the measurement value distribution of similar patients by, for example, calculating a probability distribution for the multiple measurement values ​​of similar patients extracted from the patient DB 11. For this reason, the measurement value distribution is, for example, data stored in the patient DB 11 and a probability distribution calculated from that data. The identification function 14 outputs the measurement values ​​of similar patients when they were healthy and when they developed a disease.

[0014] The detection function 15 detects a trend change in the biological information at a predetermined timing based on time series data related to the biological information.

[0015] The first calculation function 16 receives as input the patient's biometric information acquired by the acquisition function 13 and the measurement value distribution of similar patients (biometric information related to similar patients) in a certain state (one or more types) when healthy identified by the identification function 14. The first calculation function 16 calculates a first propensity score that represents the similarity in tendency between the patient's biometric information and the measurement value distribution of similar patient information when healthy. The first calculation function 16 outputs the calculated first propensity score to the discrimination function 18.

[0016] The second calculation function 17 receives as input the patient's biometric information acquired by the acquisition function 13 and the measurement value distribution of similar patients in a certain state (one or more types) at the time of disease onset identified by the identification function 14. The second calculation function 17 calculates a second propensity score that represents the similarity in tendency between the patient's biometric information and the measurement value distribution of similar patients at the time of disease. The second calculation function 17 outputs the calculated second propensity score to the discrimination function 18.

[0017] The discrimination function 18 compares the first propensity score output by the first calculation function 16 with the second propensity score output by the second calculation function 17, and outputs the comparison result to the output interface 19.

[0018] The output interface 19 outputs the comparison result output by the discrimination function 18 to the presentation device 2. The output interface 19 may be equipped with a wired or wireless communication function.

[0019] The storage unit 20 stores thresholds, formulas, programs for performing processing, and the like used in the processing of the medical information processing device 1.

[0020] 1 is merely an example and is not limiting. For example, the patient DB 11 may be connected to the medical information processing device 1 via a network. The medical information processing device 1 may also include an input interface to which input devices such as a keyboard and a mouse are connected.

[0021] In this specification, the input interface is not limited to an interface having physical operation parts such as a mouse, keyboard, etc. For example, an example of an input interface also includes 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 this electrical signal to a control circuit.

[0022] In addition, the processing circuit 12 realizes the functions of an acquisition function 13, an identification function 14, a detection function 15, a first calculation function 16, a second calculation function 17, and a discrimination function 18, for example, by a hardware processor executing a program stored in the memory unit 20.

[0023] The hardware processor refers to a circuit such as a central processing unit (CPU), a graphics processing unit (GPU), an application specific integrated circuit (ASIC), or a programmable logic device (e.g., a simple programmable logic device (SPLD) or a complex programmable logic device (CPLD), or a field programmable gate array (FPGA)). Instead of storing a program in the memory unit 20, the program may be directly embedded in the circuit of the hardware processor. In this case, the hardware processor realizes its function by reading and executing the program embedded in the circuit. The hardware processor is not limited to a single circuit, but may be configured as a single hardware processor by combining multiple independent circuits to realize each function. Furthermore, multiple components may be integrated into a single hardware processor to realize each function.

[0024] [Example of information stored in the patient database] Next, an example of information stored in the patient DB 11 will be described. FIG. 2 is a diagram showing an example of information stored in the patient DB. As shown in FIG. 2, the patient DB 11 stores, for example, patient identification information, attribute information, and measurement values ​​in association with each other. As described above, the attribute information is, for example, disease name, age, sex, etc. Measurement values ​​are, for example, blood pressure, body temperature, and pulse rate, and each is associated with a label indicating whether the measurement is in a healthy state or in a sick state. Note that the example shown in FIG. 2 is merely an example and is not limiting.

[0025] [Processing performed by specific functions] Next, we will further explain the processing performed by the identification function 14. The identification function 14 identifies similar patients from the patient DB 11 based on the attribute information of the target patient, and extracts the measurement value distribution of the identified similar patients. Note that if there are no patients with the same disease, the identification function 14 may extract a patient with the closest attribute information or the target patient itself.

[0026] FIG. 3 is a diagram illustrating an example of identifying a similar patient from the patient DB. As shown by the dashed line g11 in FIG. 3, the patient's attribute information indicates that the disease is myocardial infarction, the patient is 84 years old, and the patient is male. In this case, the identification function 14 identifies patient C from the patient DB 11, who has the same underlying disease, the same gender, and an age close to that of the target patient, as shown by the dashed line g12 in FIG. 3. In this way, the identification function 14 identifies at least one patient from the patient DB 13 who has similar attributes to the patient. Note that if no patients with the same disease are stored in the patient DB 11, the identification function 14 may identify a patient of the same gender and within a predetermined age range of the target patient.

[0027] Here, the measurement values ​​(biological information) will be explained. The measurement values ​​stored in the patient DB11 are preferably those obtained in situations where the probability of measurement errors is low (for example, upon admission or outpatient visits). Furthermore, the measurement values ​​stored in the patient DB11 may be values ​​after pre-processing (for example, smoothing) to reduce measurement errors in order to suppress the influence of measurement errors. Alternatively, the specification function 14 may perform pre-processing on the measurement values ​​stored in the patient DB11 when in use. Furthermore, the measurement values ​​stored in the patient DB11 are labeled, for example, with the physical condition of the patient at the time of measurement. For example, the label is 0 for healthy (no disease) and 1 for disease.

[0028] [Example of how scores are calculated] Next, an example of a method for calculating a propensity score performed by the first calculation function 16 and the second calculation function 17 will be described with reference to Figs. 4 and 5. Fig. 4 is a diagram for explaining an example of a method for calculating a propensity score performed by the first calculation function and the second calculation function. Fig. 5 is a diagram showing an example of a propensity score calculated by the first calculation function and the second calculation function. In the examples shown in Figs. 4 and 5, the measured values ​​are blood pressure, body temperature, and pulse rate.

[0029] As shown in image g20 of FIG. 4, the first calculation function 16 and the second calculation function 17 each exclude blood pressure (g31, g41, g51) from the patient's measurement value (biometric information) (g30), the measurement value (biometric information) of a similar patient when healthy (g40), and the measurement value (biometric information) of a similar patient when ill (g50). The first calculation function 16 calculates a first propensity score Sim between the patient's body temperature and pulse rate (g32) and the body temperature and pulse rate (g42) of a similar patient when healthy. nor As a result of the calculation, the first propensity score Sim nor The second calculation function 17 calculates the second propensity score Sim between the patient's temperature and pulse (g33) and the temperature and pulse of similar patients at the time of illness (g53). abnor As a result of the calculation, the second propensity score Sim abnor was 0.2.

[0030] Next, as shown in image g60 in FIG. 4, the first calculation function 16 and the second calculation function 17 each exclude body temperature (g71, g81, g91) from the patient's measurement value (g70), the measurement value of a similar patient when healthy (g80), and the measurement value of a similar patient when ill (g90). The first calculation function 16 calculates a first propensity score Sim between the patient's blood pressure and pulse rate (g72) and the blood pressure and pulse rate of a similar patient when healthy (g82). nor The second calculation function 17 calculates a second propensity score Sim between the blood pressure and pulse rate (g73) of the target patient and the blood pressure and pulse rate (g93) of a similar patient at the time of illness. abnor Calculate.

[0031] Then, the first calculation function 16 and the second calculation function 17 each exclude the pulse from the patient's measurements, the measurements of similar patients when healthy, and the measurements of similar patients when ill. The first calculation function 16 calculates a first propensity score Sim between the patient's blood pressure and body temperature and the blood pressure and body temperature of similar patients when healthy. nor The second calculation function 17 calculates a second propensity score Sim between the patient's blood pressure and body temperature and the blood pressure and body temperature of similar patients at the time of illness. abnor Calculate.

[0032] As described above, in this embodiment, some measurement values ​​are excluded and the similarity (score) of the tendency between the patient and similar patients is calculated. This means that the similarity of the tendency is calculated after excluding feature values ​​in which measurement errors occurred. As a result, in this embodiment, it is possible to classify into classes in which measurement errors and disease are present (because information on measurement errors is not used). Because it is not known in which biological information the measurement error occurred, in this embodiment, a comprehensive search is performed while calculating the propensity score by excluding each piece of biological information one by one. Note that Figures 4 and 5 show a case in which a measurement error occurred in one piece of biological information. The number of measurement errors may be determined depending on the type of patient information. Furthermore, although the examples shown in Figures 4 and 5 show an example in which there are three pieces of biological information, the number of pieces of biological information may be, for example, two, four, or more.

[0033] The propensity score has a larger value as the tendency between the patient's biological information and the measurement values ​​of similar patients is closer, i.e., the closer the measurement values ​​are, and the value has a smaller value as the tendency between the patient's biological information and the measurement values ​​of similar patients is more distant, i.e., the more distant the measurement values ​​are. An example of a method for calculating the propensity score will be described later.

[0034] In the examples described using Figures 4 and 5, there is one example of the condition of similar patients. In the following example, an example of a method for calculating a propensity score when multiple conditions of similar patients are extracted from the patient DB 11 will be described using Figure 6. Figure 6 is a diagram for explaining a method for calculating a propensity score when multiple conditions of similar patients are extracted from the patient DB. Examples of multiple conditions in health include when a chronic disease has not yet developed, when a chronic disease has developed, and measurement values ​​that vary depending on the time of day (for example, morning and night). Examples of multiple conditions at the time of disease onset include the early stage of disease onset and the middle stage of disease onset.

[0035] Even in the same healthy or diseased state, there is a possibility that the tendency of the tendency value may differ depending on the state. For this reason, in this embodiment, in order to deal with such cases, the propensity score may be calculated for each condition as described above. In such a case, the first calculation function 16 and the second calculation function 17 each calculate a propensity score for each state. For example, as shown in FIG. 6, the first calculation function 16, after excluding blood pressure, calculates the first propensity score Sim between the patient's biological information and the measured values ​​of similar patients when healthy in the first state nor_0. nor , and the first propensity score Sim between the patient's biological information and the measurements of similar patients in a healthy state in the second state nor_1 is calculated. nor After excluding the blood pressure, the second calculation function 17 calculates a second propensity score Sim between the patient's biological information and the measured values ​​of similar patients at the time of illness in the first state nor_0. abnor Calculate the second propensity score Sim between the patient's biological information and the measurements of similar patients at the time of the disease in the second state nor_1. abnor Calculate.

[0036] Next, an example of how to calculate the propensity score value will be described with reference to FIGS. 7 and 8. FIG. 7 is a diagram for explaining how to calculate the propensity score by principal component analysis. The first calculation method is a method using the principal component analysis technique. In the example of FIG. 7, a plurality of similar patients to the patient are extracted from the patient DB 11. The principal components are blood pressure, body temperature, pulse rate, etc. As described above, the principal component analysis is performed on the remaining measured values ​​after excluding one measured value. In such a case, the first calculation function 16 and the second calculation function 17 each analyze the patient's biometric information and the measured values ​​of the similar patients using the principal component analysis technique, and calculate the distance between the center position (origin) of the distribution of similar patients and the position of the patient, thereby calculating the propensity score. The first calculation function 16 and the second calculation function 17 each calculate the propensity score by, for example, using T as in the following equation (1): 2 The propensity score is calculated by calculating the inverse of the statistic (the square of the distance from the origin in principal component analysis). Note that in formula (1), i represents the type of biological information (for example, blood pressure i=0, body temperature i=1, pulse rate i=2).

[0037]

number

[0038] In the second calculation method, the propensity score is calculated using a maximum likelihood estimation technique. FIG. 8 is a diagram for explaining how to obtain a propensity score by maximum likelihood estimation. In this case, the first calculation function 16 and the second calculation function 17 each calculate the likelihood (P(X|Θ)) of the target patient in the parameter (Θ) obtained by maximum likelihood estimation (P(X|Θ)) for similar patients, as shown in FIG. * |Θ)) to calculate the propensity score. For example, in FIG. 8, point g121 is the patient's biological information, and line g122 is the measurement value of a similar patient. The first calculation function 16 calculates the first propensity score by calculating the likelihood of the target patient in parameters obtained by maximum likelihood estimation for similar patients when healthy. The second calculation function 17 calculates the second propensity score by calculating the likelihood of the target patient in parameters obtained by maximum likelihood estimation for similar patients when ill.

[0039] The first calculation function 16 and the second calculation function 17 may each calculate the measurement value distribution and probability distribution of similar patients in Figures 7 and 8 for each similar patient if there are many measurements for one similar patient. Alternatively, the first calculation function 16 and the second calculation function 17 may each calculate the measurement value distribution and probability distribution of similar patients using measurements for multiple similar patients. In the first calculation method or the second calculation method, the first calculation function 16 and the second calculation function 17 each calculate the propensity score using biometric information acquired about the target patient. For this reason, the biometric information of the target patient is, for example, a measurement value measured at a predetermined timing (e.g., morning) from data acquired in a time series at predetermined intervals.

[0040] 7 and 8 are merely examples of the calculation method of the propensity score, and the present invention is not limited to these. Each of the first calculation function 16 and the second calculation function 17 may calculate the propensity score using, for example, another statistical method.

[0041] [Processing performed by the discrimination function] Next, the processing performed by the discrimination function 18 will be described. FIG. 9 is a diagram for explaining the processing performed by the discrimination function. Table g200 shows possible cases for the patient's biometric information. As shown in table g200, the possible cases are: Case I: no measurement error and no disease onset; Case II: measurement error and no disease onset; Case III: no measurement error and disease onset; and Case II: measurement error and disease onset. Note that the temporal changes in the patient's biometric information shown in FIG. 9 are images for each case, and the measured values ​​used in the propensity score and comparison are not measured over time but are, for example, one value.

[0042] Graphs g211 and g212 enclosed in a rectangle g210 show examples of measurement values ​​of similar patients. Graph g211 shows the measurement values ​​of a similar patient when healthy, and graph g212 shows the measurement values ​​of a similar patient when the patient developed a disease. In graphs g211 and g212, the horizontal axis represents time, and the vertical axis represents the measurement values ​​(blood pressure, body temperature, pulse rate).

[0043] Graphs g221 to g224 enclosed by a rectangle g220 represent examples of patient biometric information for each case. Graph g221 is the biometric information of the patient in Case I, graph g222 is the biometric information of the patient in Case II, graph g223 is the biometric information of the patient in Case III, and graph g224 is the biometric information of the patient in Case IV. In graphs g221 to g224, the horizontal axis represents time, and the vertical axis represents measured values ​​(blood pressure, body temperature, pulse rate).

[0044] The discrimination function 18 compares the calculated propensity scores to estimate the presence or absence of disease onset and measurement error. In this case, the discrimination function 18 estimates which case in table g200 it corresponds to by assuming a case that is likely to be relevant among the assumed cases described above.

[0045] For example, in case I, i.e., when there is no measurement error and no disease onset, the patient's biological information is expected to be close to the measurements of similar patients when healthy. In case III, i.e., when there is no measurement error and the disease onset, the patient's biological information is expected to be close to the measurements of similar patients when ill. In case II, i.e., when there is a measurement error and the disease onset, the patient's biological information is close to the measurements of similar patients when healthy, but the patient's biological information is expected to have changed or is different. In case IV, i.e., when there is a measurement error and the disease onset, the patient's biological information is close to the measurements of similar patients when ill, but the patient's biological information is expected to have changed or is different.

[0046] For this reason, in this embodiment, the discrimination function 18 classifies the above cases based on the magnitude relationship between the first propensity score of the patient's biometric information and the measurements of similar patients when healthy, and the second propensity score of the patient's biometric information and the measurements of similar patients when ill, as well as the magnitude relationship between the difference (or ratio) and a threshold value.

[0047] 10 is a flowchart showing an example of the calculation and discrimination process of the propensity score. Each of the first calculation function 16 and the second calculation function 17 initializes n (for example, to 0) (step S101). Each of the first calculation function 16 and the second calculation function 17 calculates the propensity score by excluding the n-th measurement value (step S102). Each of the first calculation function 16 and the second calculation function 17 determines whether n and m are equal (step S102). If n and m are not equal, each of the first calculation function 16 and the second calculation function 17 adds 1 to n (step S104) and returns to the process of step S102.

[0048] If n and m are equal, the discrimination function 18 acquires the first propensity score and the second propensity score (step S105). The discrimination function 18 compares the first propensity score with the second propensity score (step S106). Based on the comparison result, the discrimination function 18 classifies the case into three cases: Case I, Case II, Case III, and Case IV (step S107).

[0049] FIG. 11 is a diagram showing an example of a first propensity score and an example of a second propensity score for each case. In FIG. 11, a large propensity score is, for example, 0.8 or more in the normalized score, a medium propensity score is, for example, 0.7 to 0.4 in the normalized score, and a small propensity score is, for example, 0.3 or less in the normalized score. In the example of FIG. 11, Case I corresponds to the first propensity score Sim when blood pressure is excluded. nor is large and the second propensity score Sim abnor is small, and the first propensity score Sim nor is large and the second propensity score Sim abnor is small, and the first propensity score Sim nor is large and the second propensity score Sim abnor is small.

[0050] Case II corresponds to the first propensity score Sim nor is large and the second propensity score Sim abnoris small, and the first propensity score Sim nor is medium and the second propensity score Sim abnor is small, and the first propensity score Sim nor is medium and the second propensity score Sim abnor is small.

[0051] Case III corresponds to the first propensity score Sim nor is small and the second propensity score Sim abnor is large, and the first propensity score Sim nor is small and the second propensity score Sim abnor is large, and the first propensity score Sim nor is small and the second propensity score Sim abnor is large.

[0052] Case IV corresponds to the first propensity score Sim nor is small and the second propensity score Sim abnor is large, and the first propensity score Sim nor is small and the second propensity score Sim abnor The first propensity score Sim when excluding pulse is medium. nor is small and the second propensity score Sim abnor is in the middle.

[0053] In this embodiment, the discrimination function 18 classifies cases into three categories: Case I, Case II, Case III, and Case IV, based on the trends shown in Fig. 11. Note that the example of time-varying measured values ​​shown in Fig. 9 and the example of large / medium / small relationship of propensity scores for each case shown in Fig. 11 are merely examples, and are not limited to these.

[0054] [Example of processing by medical information processing device] Next, a description will be given of an example of a processing procedure performed by the medical image processing apparatus 1. Fig. 12 is a flowchart of an example of a processing procedure performed by the medical image processing apparatus according to this embodiment.

[0055] The acquisition function 13 acquires patient information (biometric information, attribute information) of the patient (step S201).

[0056] The identifying function 14 identifies similar patients whose attribute information is similar to that of the patient from the patient DB 11 based on the acquired attribute information of the patient (step S202).

[0057] The identifying function 14 calculates the distribution of measured values ​​of similar patients in a healthy state using the patient information of the identified similar patients (step S211). The first calculating function 16 calculates a first propensity score Sim between the patient's biological information and the distribution of measured values ​​of similar patients in a healthy state. nor is calculated (step S212).

[0058] The identifying function 14 calculates the distribution of measured values ​​of similar patients at the time of the disease using the patient information of the identified similar patients (step S221). The second calculating function 17 calculates a second propensity score Sim between the patient's biological information and the distribution of measured values ​​of similar patients at the time of the disease. abnor is calculated (step S222).

[0059] The discrimination function 18 is the first propensity score Sim nor is the second propensity score Sim abnor The discrimination function 18 determines whether the first propensity score Sim nor is the second propensity score Sim abnor If it is greater than 1, the first propensity score Sim nor and the second propensity score Sim abnor It is determined whether the absolute value of the difference between the two is greater than a threshold value (step S232).

[0060] First propensity score Sim nor is the second propensity score Sim abnor Larger than the first propensity score Sim nor and the second propensity score Sim abnor If the absolute value of the difference is greater than the threshold value, the discrimination function 18 determines that the subject patient has "normal physical condition" because there are no measurement errors in the measured values ​​and no disease onset (Case I) (Step S233).

[0061] First propensity score Sim nor is the second propensity score Sim abnor Larger than the first propensity score Sim nor and the second propensity score Sim abnor If the absolute value of the difference is equal to or less than the threshold value, the discrimination function 18 determines that the subject patient has a measurement error and no disease has developed (Case II), and therefore determines that the subject patient is in "normal physical condition and the measurement is abnormal" (Step S234).

[0062] Second propensity score Sim abnor is the first propensity score Sim nor In the above cases, the discrimination function 18 determines that a disease has developed (case III or IV) and therefore determines that the disease has developed (step S235).

[0063] The reason why the discrimination function 18 judges that a disease has developed when a disease has developed regardless of whether there is a measurement error or not is because it is important to inform the user (e.g., a doctor, nurse, caregiver, etc.) that a disease has developed.

[0064] 12 is an example and is not limiting. For example, the medical information processing apparatus 1 may perform the processes of steps S211 to S212 and the processes of steps S221 to S222 simultaneously or in a time-division manner, may perform the processes of steps S211 to S212 after the processes of steps S221 to S222, or may perform the processes of steps S211 to S212 after the processes of steps S221 to S222.

[0065] [Example] Next, an example of a usage scenario of the medical information processing device 1 will be described. FIG. 13 is a diagram illustrating an example of a usage scenario of the medical information processing device. In the following embodiment, as shown in FIG. 13, the medical information processing device 1 determines whether a measurement error or disease has occurred based on vital signs measured by a patient (target patient) at home, and shares the information with a doctor in the hospital. Note that the measuring device used by the target patient for measurement at home, etc., may have, for example, a wireless communication function, and the measuring device may transmit biological information to the medical information processing device 1 via a network NW. Alternatively, the target patient or a family member of the target patient may input the biological information into a terminal (e.g., a personal computer, tablet terminal, smartphone, etc.) in the target patient's home, and the terminal may transmit the biological information to the medical information processing device 1 via the network NW. Alternatively, the terminal may acquire biological information through short-range communication between the measuring device and the terminal, and transmit the acquired biological information to the medical information processing device 1 via the network NW. Note that the measuring device may be a wearable terminal, such as a smart watch, equipped with a sensor capable of detecting biological information and a terminal function.

[0066] FIG. 14 is a diagram showing an example of information presented when it is determined that there are no measurement errors and no disease onset. FIG. 14 shows a case where the medical information processing device determines that the biological information is Case I, and is an example of image g300 displayed, for example, on an image display device of a personal computer used by a doctor or a tablet terminal. Image g300 includes, for example, a patient ID (patient identification information) image g301 and an image g302 showing the time change of the measurement values ​​included in the patient information. In this case, the medical information processing device 1 may be configured to present only the patient information as shown in FIG. 14. Note that the medical information processing device 1 may also be configured to determine, for example, "no measurement errors and no disease onset" and present an image g303 showing the determination result.

[0067] FIG. 15 is a diagram showing an example of information presented when it is determined that a measurement error occurred but no disease has developed. FIG. 15 shows a case where the medical information processing device 1 has determined that the biological information is Case II, and is an example of image g310 displayed on, for example, the image display device of a personal computer used by a doctor or a tablet terminal. Image g310 includes, for example, a patient ID image g301, an image g312 showing the time change in the measurement values ​​included in the patient information, an image g313 showing, for example, the results of principal component analysis, and an image g314 showing a message suggesting a measurement error based on the determination result (for example, "A measurement error occurred in body temperature"). Note that the medical information processing device 1 may also present an image indicating, for example, "A measurement error occurred but no disease has developed."

[0068] Fig. 16 is a diagram showing an example of information presented when it is determined that a disease has developed. Fig. 16 shows a case where the medical information processing device 1 has determined that the biological information is Case III or IV, and is an example of image g320 that is displayed on, for example, an image display device of a personal computer or a tablet terminal used by a doctor. Image g320 includes, for example, a patient ID image g301, an image g322 showing the time change in the measurement values ​​included in the patient information, an image g323 of the results of principal component analysis, and an image g324 of a message encouraging a house call based on the determination result (for example, "The patient shows similar trends to patients with myocardial infarction. A house call is required").

[0069] 14 to 16 are merely examples, and the images presented are not limited to these. The images presented may also include other information (for example, the patient's age, sex, and disease).

[0070] As described above, in this embodiment, the medical information processing device 1 is provided with an acquisition function 13 that acquires biometric information about a patient, a detection function 15 that detects trend changes in the biometric information at a predetermined timing based on time series data about the biometric information, and a discrimination function 18 that determines whether the trend change is due to a measurement error in the biometric information or a change in the patient's condition.

[0071] As a result, according to this embodiment, when an abnormality occurs in the biological information of a subject patient, it is possible to clearly distinguish whether the cause is a measurement abnormality or the onset of a disease.

[0072] In addition, this embodiment further includes an identification function 14 that identifies similar patients based on the patient's attribute information, and a discrimination function 18 that discriminates trend changes from the above-mentioned state changes based on the patient's biometric information at a predetermined timing in the biometric information record of the similar patient.

[0073] As a result, according to this embodiment, the acquisition function 13 acquires multiple types of biometric information. Then, in this embodiment, the propensity score is calculated while, for example, excluding one piece of biometric information from the acquired multiple pieces of biometric information one by one.

[0074] As a result, according to this embodiment, when a measurement error occurs in one of three pieces of biological information, for example, it is possible to appropriately determine that this is a measurement error and present it.

[0075] In the above example, the first and second propensity scores are calculated while, for example, excluding one piece of each of a plurality of pieces of biological information, and based on the calculated first and second propensity scores, it is determined whether a trend change in the biological information is due to a measurement error in the biological information or a change in the patient's condition. However, the present invention is not limited to this. The number of pieces of biological information may be one.

[0076] When there is only one piece of biometric information, the medical information processing device 1 identifies similar patients based on attribute information included in the patient's biometric information. The medical information processing device 1 calculates a first propensity score based on the patient's biometric information and measurements of similar patients when they were healthy, and calculates a second propensity score based on the patient's biometric information and measurements of similar patients when they were ill. The medical information processing device 1 calculates the propensity score using, for example, the maximum likelihood estimation technique, which is the second calculation method described above. The medical information processing device 1 may determine whether a trend change in the biometric information is due to a measurement error in the biometric information or a change in the patient's condition, based on the first propensity score and the second propensity score.

[0077] Thus, even if there is only one piece of biological information, according to this embodiment, when an abnormality occurs in the biological information of the subject patient, it is possible to clearly distinguish whether the cause is a measurement abnormality or the onset of a disease.

[0078] [Variations] 15 and 16, the medical information processing device 1 determines whether the case is, for example, Case I, Case II, or Case III or IV, and presents the determination result, but the present invention is not limited to this. For example, the medical information processing device may detect data in which a measurement error has occurred and screen the learning data as a preprocessing step before loading the data into a CDS (Clinical Decision Support) model.

[0079] Fig. 17 is a diagram showing an example of the configuration of a modified medical information processing device. As shown in Fig. 17, the medical information processing device 1A includes a patient DB 11, a processing circuit 12A, an output interface 19, a memory unit 20, and a learning model storage unit 22. The processing circuit 12A includes, for example, an acquisition function 13, a specification function 14, a detection function 15, a discrimination function 18, and an extraction function 21. The detection function 15 includes a first calculation function 16 and a second calculation function 17. The learning model storage unit 22 includes a learning model 23.

[0080] The learning model 23 is, for example, a model for CDS. The CDS model may be one used in clinical practice.

[0081] The extraction function 21 performs, for example, moving average processing for each sampling time and type of biological information (for example, body temperature, respiratory rate, pulse rate) on the biological information acquired by the acquisition function 13, and extracts and counts biological information that differs by a predetermined value or more from the moving average value. The sampling time is, for example, every second.

[0082] In addition, the processing circuit 12A realizes the functions of the acquisition function 13, the identification function 14, the detection function 15, the first calculation function 16, the second calculation function 17, the discrimination function 18, and the extraction function 21, for example, by the hardware processor executing a program stored in the memory unit 20.

[0083] FIG. 18 is a diagram showing an example of an image presented on the presentation device. Image g330 includes, for example, a patient ID image g301, an image g332 showing temporal changes in biometric information (measurements) included in the patient information, and a screening result image g333 for the measurements. In image g332, measurement values ​​surrounded by ellipses g341 to g344 are examples of measurement values ​​detected by the extraction function 21 as measurement values ​​having a difference of at least a predetermined value. Note that the biometric information in the example of FIG. 18 may be, for example, measurement values ​​acquired at each sampling time. Image g333 includes, for example, training data items and model accuracy items. The training data items include, for example, item and no screening. The items include screening, number of samples, mean value, and variance. The model accuracy item includes a value representing the model accuracy before and after additional learning using data without screening, and a value representing the model accuracy before and after additional learning using data with screening. Note that image g333 is an example and is not limited to this, and other items may be included.

[0084] According to this embodiment, by presenting such image g330 to the user, it is possible to provide the user with information to determine whether or not to use the acquired measurement values ​​for training the learning model 23. Note that the user may select whether to use the measurement values ​​with screening or the measurement values ​​without screening for training the learning model 23.

[0085] The medical information processing device 1A may select whether to use a measurement value with screening or a measurement value without screening for training the training model 23 based on, for example, at least one of the variance value and the model accuracy. In this case, the medical information processing device 1A may select to use a measurement value without screening if, for example, the variance value is within a threshold value. Alternatively, the medical information processing device 1A may select to use a measurement value without screening if, for example, the difference or ratio between the values ​​of model accuracy before and after training is within a threshold value.

[0086] The first calculation function 16, the second calculation function 17, and the discrimination function 18 may use the learning model 23 thus learned to calculate the propensity score or to discriminate by comparing the propensity scores.

[0087] In accordance with at least one embodiment described above, having processing circuitry 12 (or 12A) allows for differentiation between measurement errors and disease onset based on the proximity of trends to other patients.

[0088] In the above-described embodiment, the acquisition function 13 is an example of an "acquisition unit", the detection function 15 is an example of a "detection unit", the discrimination function 18 is an example of a "discrimination unit", the identification function 14 is an example of a "identification unit", and the extraction function 21 is an example of an "extraction unit".

[0089] 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 forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their modifications are included within the scope and spirit of the invention, as well as within the scope of the invention and its equivalents as defined in the claims. [Explanation of symbols]

[0090] 1, 1A...Medical information processing device, 11...Past patient DB, 12, 12A...processing circuit, 13...acquisition function, 14...Specific functions, 15...detection function, 16...First calculation function, 17...second calculation function, 18...Discrimination function, 19...output interface, 20...Storage section, 21...Extraction function, 22...Model storage section, 23...Learning Model

Claims

1. an acquisition unit for acquiring biometric information about a patient; a detection unit that detects a trend change in the biological information at a predetermined timing based on time-series data related to the biological information; a determination unit that determines whether the trend change is due to a measurement error of the biological information or a change in the condition of the patient; an identification unit that identifies similar patients based on the attribute information of the patients; Equipped with the acquisition unit acquires a plurality of types of biometric information, the detection unit detects a trend change in the biometric information at a predetermined timing, excluding any one of the plurality of types of acquired biometric information; calculates a first propensity score based on the biometric information of the patient at the predetermined timing and biometric information of similar patients when healthy; and calculates a second propensity score based on the biometric information of the patient at the predetermined timing and biometric information of the similar patients when sick; the determination unit determines whether the trend change is due to a measurement error of the biological information or a change in the condition of the patient based on the first propensity score and the second propensity score; Medical information processing equipment.

2. The detection unit calculating the first propensity score based on a result of performing a principal component analysis on the biological information of the patient at the predetermined timing and biological information of the similar patient in a healthy state; calculating the second propensity score based on a result of performing a principal component analysis on the biological information of the patient at the predetermined timing and biological information of the similar patient at the time of illness; The medical information processing device according to claim 1 .

3. The detection unit calculating the first propensity score based on an estimated likelihood of the biological information of the patient at the predetermined timing and an estimated likelihood of biological information of the similar patient in a healthy state; calculating the second propensity score based on an estimated likelihood of the biological information of the patient at the predetermined timing and an estimated likelihood of biological information of the similar patient at the time of the disease; The medical information processing device according to claim 1 .

4. An extraction unit that extracts data in which the measurement error occurred from the biological information of the patient in whom the judgment unit has judged that the trend change is due to a measurement error; an output interface that presents information about the extracted data; Further provided with The medical information processing device according to any one of claims 1 to 3.

5. A medical information processing device Obtaining biometric information about the patient; Detecting a trend change in the biological information at a predetermined timing based on time-series data relating to the biological information; determining whether the trend change is due to a measurement error of the vital sign or a change in the condition of the patient; Identifying similar patients based on the attribute information of the patients; In the acquisition of the biometric information, a plurality of types of the biometric information are acquired, In detecting the trend change, any one of the acquired types of biometric information is excluded, and a trend change of the biometric information at a predetermined timing is detected; a first propensity score is calculated based on the biometric information of the patient at the predetermined timing and biometric information of similar patients when healthy; and a second propensity score is calculated based on the biometric information of the patient at the predetermined timing and biometric information of the similar patients when sick; In determining the trend change, it is determined whether the trend change is due to a measurement error of the biological information or a change in the condition of the patient, based on the first propensity score and the second propensity score. Medical information processing method.

6. On the computer, obtaining biometric information about the patient; detecting a trend change in the biological information at a predetermined timing based on time-series data relating to the biological information; determining whether the change in trend is due to a measurement error of the vital signs or a change in the condition of the patient; Identifying similar patients based on the attribute information of the patients; In the acquisition of the biometric information, a plurality of types of biometric information are acquired; In detecting the trend change, any one of the acquired types of biometric information is excluded, and a trend change of the biometric information at a predetermined timing is detected; a first propensity score is calculated based on the biometric information of the patient at the predetermined timing and biometric information of similar patients when healthy; and a second propensity score is calculated based on the biometric information of the patient at the predetermined timing and biometric information of the similar patients when sick; In determining the trend change, it is determined whether the trend change is due to a measurement error of the biological information or a change in the condition of the patient, based on the first propensity score and the second propensity score. program.

7. an acquisition unit for acquiring biometric information about a patient; a detection unit that detects a trend change in the biological information at a predetermined timing based on time-series data related to the biological information; a determination unit that determines whether the trend change is due to a measurement error of the biological information or a change in the condition of the patient; an identification unit that identifies similar patients based on the attribute information of the patients; Equipped with The detection unit calculating a first propensity score based on the biological information of the patient at the predetermined timing and biological information of similar patients in a healthy state; calculating a second propensity score based on the biological information of the patient at the predetermined timing and biological information of the similar patient at the time of the disease; the determination unit determines whether the trend change is due to a measurement error of the biological information or a change in the condition of the patient based on the first propensity score and the second propensity score; Medical information processing equipment.

8. A medical information processing device Obtaining biometric information about the patient; Detecting a trend change in the biological information at a predetermined timing based on time-series data relating to the biological information; determining whether the trend change is due to a measurement error of the vital sign or a change in the condition of the patient; Identifying similar patients based on the attribute information of the patients; The trend change detection includes: calculating a first propensity score based on the biological information of the patient at the predetermined timing and biological information of similar patients in a healthy state; calculating a second propensity score based on the biological information of the patient at the predetermined timing and biological information of the similar patient at the time of the disease; In determining the trend change, it is determined whether the trend change is due to a measurement error of the biological information or a change in the condition of the patient, based on the first propensity score and the second propensity score. Medical information processing method.

9. On the computer, obtaining biometric information about the patient; detecting a trend change in the biological information at a predetermined timing based on time-series data relating to the biological information; determining whether the change in trend is due to a measurement error of the vital signs or a change in the condition of the patient; Identifying similar patients based on the attribute information of the patients; The trend change detection includes: calculating a first propensity score based on the biological information of the patient at the predetermined timing and biological information of similar patients in a healthy state; calculating a second propensity score based on the biological information of the patient at the predetermined timing and biological information of the similar patient at the time of illness; In determining the trend change, it is determined whether the trend change is due to a measurement error of the biological information or a change in the condition of the patient, based on the first propensity score and the second propensity score. program.

Images