The type designation unit 213 designates the type of physiological parameter for the measurement information and prediction information acquired by the acquisition unit 211. The type designation unit 213 designates the type of physiological parameter, for example, based on an instruction from the user. A user inputs an instruction via the input 250, for example. The acquisition unit 211 may acquire the measurement information and the prediction information regarding the type of physiological parameter designated by the type designation unit 213. By designating the type of physiological parameter, it becomes possible to perform an analysis more suited to the user's application.

The calculation unit 214 calculates the degree of divergence between the prediction information and the measurement information acquired by the acquisition unit 211. The degree of divergence is an index representing the accuracy of the prediction of a patient's physiological parameter by a medical practitioner. A large degree of divergence indicates low accuracy of prediction, and a small degree of divergence indicates high accuracy of prediction. The degree of divergence is, for example, the difference between prediction information and measurement information. When a plurality of differences can be calculated, the degree of divergence may be the median value, average value, or the like of the plurality of differences. The calculation unit 214 calculates, for example, the degree of divergence between the prediction information and the measurement information regarding the physiological parameter of the type designated by the type designation unit 213.

The calculation unit 214 calculates, for example, the degree of divergence between the prediction information at a predetermined prediction date and time and the measurement information measured at this prediction date and time. For example, the calculation unit 214 calculates the difference (+8) between the predicted value 110 of ART at 7:00 on September 1st at the prediction date and time and the measured value 102 of ART measured at 7:00 on September 1st (see FIG. 4). As described above, when the acquisition unit 211 acquires the first prediction information and the second prediction information, and the first measurement information and the second measurement information, the calculation unit 214 calculates a first degree of divergence between the first prediction information and the first measurement information and a second degree of divergence between the second prediction information and the second measurement information.

The calculation unit 214 may calculate the degree of divergence by normalizing each physiological parameter. The calculation unit 214 may weigh each physiological parameter and calculate a composite degree of divergence in which the degree of divergence of a plurality of types of physiological parameters is taken into consideration. This composite degree of divergence is calculated, for example, as follows. First, after setting the weight of the vector space for each physiological parameter, the single degree of divergence of each physiological parameter axis is calculated, and a point on the vector space is obtained from the single degree of divergence. The vector length from this point to the origin is calculated as the composite degree of divergence.

The central monitor 200 may have a function of notifying when the magnitude of the degree of divergence calculated by the calculation unit 214 exceeds a predetermined value.

The function is switched on or off by user setting, for example, and this ON and OFF can be set for each physiological parameter, for example. This allows the user to set alerts only for those parameters the user wants to pay particular attention to. Further, when the magnitude of the degree of divergence of at least one physiological parameter exceeds a predetermined value, a notification may be made. Such a notification function makes it easier for the medical practitioner to notice the divergence between the prediction information and the measurement information, so that the medical practitioner can quickly take measures such as changing the treatment plan. The magnitude and threshold of the degree of divergence to be notified are set for each physiological parameter, for example.

The notification may be made, for example, by displaying a warning message on the display 240, or by changing the display method of the degree of divergence (for example, changing the color of the numerical value, changing the background color, and the like). The notification may be made by sending a warning e-mail to the terminal of the person concerned, or by emitting an alarm sound from a speaker or the like.

The analyzing unit 215 analyzes the relationship between the subject information acquired by the acquisition unit 211 and the degree of divergence calculated by the calculation unit 214. Specifically, the analyzing unit 215 analyzes the relationship between the patient information and medical practitioner information included in the subject information and the magnitude of the degree of divergence. The analyzing unit 215 analyzes the relationship between the subject information and the degree of divergence for each category of the subject information classified by the classification unit 212, for example. This makes it possible to analyze the relationship between each category of patients or medical practitioners and prediction accuracy. When the calculation unit 214 calculates a first degree of divergence and the second degree of divergence, the analyzing unit 215 analyzes the relationship between the subject information and the first degree of divergence and analyzes the relationship between the subject information and the second degree of divergence. This makes it possible to analyze the relationship between subject information and the accuracy of short-term prediction and long-term prediction.

The output 216 outputs analysis information representing the relationship between the subject information and the degree of divergence, which was analyzed by the analyzing unit 215. The output 216 outputs analysis information for each category classified by the classification unit 212, for example. By checking the analysis information, a user such as a medical practitioner can grasp the degree of prediction accuracy for each category to which the subject information belongs. The analysis information includes, for example, information on the confidence interval of the degree of divergence (for example, FIG. 6A described later).

A confidence interval is, for example, a 95% confidence interval. The analysis information may include information on graded evaluation according to the degree of divergence, that is, prediction accuracy (for example, FIG. 6B described later).

FIGS. 6A and 6B show an example of the analysis information output by the output 216. FIGS. 6A and 6B show the prediction accuracies of medical practitioners in categories 2 and 3 for physiological parameters of patients in category 1. The types of predicted physiological parameters are ART, CVP, PaO2, and PaCO2. FIG. 6A shows the degree of divergence between the prediction information and the measurement information 24 hours and 72 hours after the reference date and time for these physiological parameters. In FIG. 6A, the value outside the parentheses is the median value of the difference between the predicted information and the measurement information, and the value inside the parentheses is the 95% confidence interval of this difference. In FIG. 6B, evaluations A and B are graded evaluations according to the degree of divergence, and evaluation A has a smaller degree of divergence than evaluation B, that is, the prediction accuracy is high. The graded evaluation is made, for example, in five grades. Evaluations are determined, for example, based on median dissociation and 95% confidence intervals.

Category 1 of the patients and categories 2 and 3 of the medical practitioners are, for example, as follows.

(Category 1 of Patients)

- 20 Age: Over 65 years old
- Sex: Male
- Reason for admission to intensive care unit, or the like: After heart surgery
- Medication status: Prescribed drug
- Underlying disease: High blood pressure
  
  (Category 2 of Medical Practitioners)
- Clinical department: Cardiac surgeon
- Years of experience: 10 years or more
  
  (Category 3 of Medical Practitioners)
- Clinical department: Intensive care physician
- Years of experience: 10 years or more.

In the central monitor 200 of the present embodiment, the analysis information representing the relationship between the subject information and the magnitude of the degree of divergence is output in this way, so that a user such as a medical practitioner can understand the relationship between at least one of the patient and the medical practitioner and the degree of divergence. Although the details will be described later, this makes it possible to grasp the tendency of the prediction accuracy of the physiological parameter.

The central monitor 200 may issue a notification when the degree of divergence is large in the output analysis information. The above notification may be made by, for example, displaying a warning message on the display 240, or changing the display method of the degree of divergence or category (for example, changing the color of the numerical value, changing the background color, and the like). The magnitude and threshold of the degree of the divergence to be notified are set for each physiological parameter, for example.

[Processing Method of Central Monitor 200]

Next, the analysis processing of subject information, measurement information, and prediction information by the central monitor 200, that is, the information processing method by the central monitor 200 will be described using FIG. 7 together with FIG. 6A.

FIG. 7 is a flowchart illustrating the analysis processing of the central monitor 200. The flowchart can be executed by the control unit 210 of the central monitor 200 according to a program.

First, the control unit 210 acquires subject information, measurement information, and prediction information (steps S101 and S102). The control unit 210 may acquire the subject information, the measurement information, and the prediction information at the same time, or may acquire the subject information after acquiring the measurement information and the prediction information.

Next, the control unit 210 receives the designation of the type of physiological parameters (step S103). For example, the control unit 210 receives a designation of ART, CVP, PaO2, and PaCO2 from the user via the input 250 (FIG. 6A).

Next, the control unit 210 calculates the degree of divergence between the measurement information and the prediction information acquired in step S102 for the type of physiological parameter designated in step S103 (step S104). For example, the control unit 210 subtracts the measured values of ART 24 hours and 72 hours after the reference date and time from the predicted values of ART 24 hours and 72 hours after the reference date and time, respectively, and calculates the difference therebetween (FIG. 6A). When the magnitude of the degree of divergence exceeds a predetermined value, the control unit 210 notifies using the display 240, for example.

Next, the control unit 210 determines a plurality of categories and classifies the subject information into the plurality of categories (step S105). For example, the control unit 210 classifies the patient information according to age, sex, reason for admission to an intensive care unit and the like, and medication status, based on instructions from the user via the input 250, and classifies medical practitioner information according to clinical department and years of experience.

Next, the control unit 210 analyzes the relationship between the subject information and the degree of divergence for each category classified in step S105 (step S106). The degree of divergence calculated in step S104 is used. For example, the control unit 210 obtains and analyzes the median value and the 95% confidence interval of the degree of divergence for each category (FIG. 6A).

After that, the control unit 210 outputs the analysis information analyzed in step S106 (step S107) and ends the process. The control unit 210 outputs the analysis information by causing the display 240 to display the screen illustrated in FIG. 6A, for example.

[Effects of Central Monitor 200 and Information Processing System 1]

In the central monitor 200 and the information processing system 1 of the present embodiment, since the analysis information representing the relationship between the subject information and the magnitude of the degree of divergence is output in this way, a user such as a medical practitioner can check the relationship between at least one of the patient and the medical practitioner, and the degree of divergence. Therefore, it is possible to grasp the tendency of the prediction accuracy of the physiological parameter. This effect will be described below.

For example, when a 72-year-old male patient after heart bypass surgery, i.e., a patient belonging to category 1 of the patient, was admitted to the intensive care unit and there is a disagreement about the date of leaving the intensive care unit between the attending physician (cardiac surgeon) and the intensive care physician, a user such as a medical practitioner can use the analysis information illustrated in FIG. 6A. For example, the attending physician proposes that the patient be discharged from the intensive care unit 24 hours after the reference date and time, and the intensivist care physician proposes that the patient's condition be observed while performing respiratory management in the intensive care unit for 48 hours or more from the reference date and time.

From the analysis information illustrated in FIG. 6A, the user can check the following. For the prediction of ART, cardiac surgeons (category 2) tended to predict low, intensive care physicians (category 3) tended to predict higher, and the variability of predictions (95% confidence interval) was lower for cardiac surgeons. CVP predictions showed no difference in the degree of divergence and variability between cardiac surgeons and intensive care physicians 24 hours after the reference date and time, but the degree of divergence and the variability were greater for cardiac surgeons 72 hours after the reference date and time. The variability of PaO2 predictions is wide for both cardiac surgeons and intensive care physicians 24 hours after the reference date and time, but the variability becomes less 72 hours after the reference date and time. Here, intensive care physicians show slightly less variability. For the prediction of PaCO2, cardiac surgeons had a greater degree of divergence and variability than intensive care physicians 24 hours after the reference date and time, but no difference between cardiac surgeons and intensive care physicians is seen 72 hours after the reference date and time. The central monitor 200 notifies using the display 240, for example, when the magnitude of the degree of divergence exceeds a predetermined value.

From the analysis information illustrated in FIG. 6A, it can be grasped that the prediction accuracy tends to be higher for intensive care physicians than for cardiac surgeons for respiratory conditions (for example, PaO2 and PaCO2) of patients belonging to category 1. Also, for blood pressure (ART), cardiac surgeons show less variability in predictions than intensive care physicians. Therefore, by using the analysis information, the user can easily find an appropriate policy, such as respecting the opinion of the intensive care physician regarding the respiratory management and respecting the opinion of the attending physician (cardiac surgeon) regarding the blood pressure management.

In addition, by deriving the prediction accuracy for each patient category and each medical practitioner category in this way, it becomes easier to find a patient category and a medical practitioner category with high prediction accuracy. For example, by using data of this category with high prediction accuracy as training data, it is possible to use Artificial Intelligence (AI) technology to predict the state of a patient.

In addition, the training such as points of view for changes in physiological parameters can be made to other medical practitioners from medical practitioners belonging to categories with high prediction accuracy and it is possible to effectively improve the prediction accuracy of all medical institutions. In addition, for parameters belonging to the category with low prediction accuracy (for example, medical practitioners), reports on prediction accuracy can be generated automatically or manually, which can be useful for staff training and the improvement of hospital protocols. In addition, it is possible to accumulate predicted values, measured values, and data on the degrees of divergence thereof, and use the accumulated data to improve the analysis algorithm and improve the prediction accuracy.

In addition, by identifying a patient category that tends to have a large degree of divergence, that is, a patient category that is difficult for medical practitioners to predict, medical practitioners can better care for their patients. For example, for patients in such a category, it is possible to prepare more carefully for sudden changes in the patient, or to assign more experienced staff (for example, nurses). In particular, the analysis information can be effectively used in medical settings such as intensive care units where patients of all ages and diseases enter.

In addition, by identifying a patient category that tends to have a smaller degree of divergence, that is, a patient category that can be predicted with relatively high accuracy by medical practitioners, it is possible, for example, to make appropriate staffing decisions within medical institutions. For example, the decision-making authority can be delegated to junior physicians for these categories of patients, and more unpredictable categories of patients can be assigned to senior physicians. Also, depending on the condition of the patient, the task may be delegated from the doctor to a nurse or a pharmacist. In addition, for patients in categories that are extremely difficult to predict, veteran nurses may be assigned, whereby a system may be established that allows assessment by multiple people.

In addition, by comparing the analysis information of each medical institution, it becomes easier to discover bias in the prediction accuracy of each medical institution, and it becomes possible to effectively use the analysis information of other institutions.

As described above, since the central monitor 200 outputs the analysis information representing the relationship between the subject information and the degree of divergence, a user such as a medical practitioner can check the relationship between at least one of the patient and the medical practitioner, and the magnitude of the degree of divergence. Therefore, it is possible to grasp the tendency of the prediction accuracy of the physiological parameter.

Modifications of the information processing system 1 described in the above embodiment will be described below. In the following, to avoid duplication of description, a detailed description of the same configuration as each configuration of the information processing system 1 described in the above embodiment will be omitted.

FIG. 8 is a flowchart illustrating another example of analysis processing by the control unit 210 described in FIG. 7 above. After analyzing the relationship between the subject information and the degree of divergence (step S205), the control unit 210 determines the category for classifying the subject information (step S206). In this respect, the information processing system 1 according to Modification 1 is different from the information processing system 1 described in the above embodiment.

First, the control unit 210 performs steps S201 to S204 in the same manner as steps S101 to S104 described with reference to FIG. 7 and calculates the degree of divergence. Next, the control unit 210 analyzes the relationship between the subject information and the degree of divergence (step S205).

Next, the control unit 210 determines a plurality of categories based on the relationship between the subject information and the degree of divergence analyzed in step S205 (step S206). In other words, the classification unit 212 determines a plurality of categories based on the relationship between the subject information and the degree of divergence, which was analyzed by the analyzing unit 215. The classification unit 212, for example, finds characteristics of the subject information having a correlation with the magnitude of the degree of divergence and determines a plurality of categories according to these characteristics. The classification unit 212 determines a plurality of categories using, for example, machine learning or statistics.

Next, the control unit 210 classifies each of the plurality of pieces of subject information into the plurality of categories determined in step S