Infection prediction device, infection prediction method, and program

Abstract

To provide an infection prediction device, an infection prediction method and a program with which it is possible to predict, including a cause of infection, a possibility of patients being infected with infectious diseases.SOLUTION: An infection prediction device 100 comprises: a data acquisition unit 10 for acquiring specimen inspection data that pertains to a patient's specimen inspection and patient state data that indicates a state of the patient; a specimen inspection data statistics unit 21 for finding the statistics of specimen inspection data; a patient state data classification unit 22 for classifying the patient state data into one of a plurality of preset categories; and a prediction unit 30 for inputting the statistics obtained from inspection data and the classification result of patient state data to a prediction model having been trained for a relationship between the statistics of specimen inspection data, the categories of patient state and infectious diseases by machine learning, and predicting a possibility of a patient having been infected with an infectious disease, on the basis of an output result from the prediction model.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to an infection prediction device and an infection prediction method for predicting the possibility of being infected with an infectious disease, and further relates to a program for realizing these.

Background Art

[0002] Conventionally, in the medical field, nosocomial infections caused by medical practices may occur, and it is important to suppress the occurrence of nosocomial infections. However, since it is extremely difficult to completely suppress infectious diseases in a hospital, which is a site where advanced medical treatment is provided, it is more important to detect patients infected with infectious diseases due to nosocomial infections at an early stage and perform appropriate treatment.

[0003] For this reason, for example, Patent Document 1 discloses a system for determining the risk of a patient having an infectious disease. The system disclosed in Patent Document 1 applies patient information input by a doctor to a risk score for risk factors such as "diabetes" and "liver disease" and a determination criterion for converting the risk score into a risk level, thereby evaluating the resistance of a patient to an infection source in five levels.

[0004] By using the system disclosed in Patent Document 1, it is possible to determine the risk that a patient who has received medical treatment will be infected due to nosocomial infection. Therefore, it is considered that patients who are likely to be infected can be intensively monitored, and the occurrence of nosocomial infection can be suppressed.

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0006] However, in the actual medical field, it is required to identify the cause of nosocomial infection. However, the system disclosed in Patent Document 1 only determines whether a patient is likely to contract a nosocomial infection, and it is impossible to identify the cause with this system.

[0007] An example of the object of the present invention is to provide an infection prediction device, an infection prediction method, and a program that can predict the possibility of a patient contracting an infectious disease, including the cause of the infection.

Means for Solving the Problems

[0008] To achieve the above object, an infection prediction device according to one aspect of the present invention includes: a data acquisition unit that acquires specimen test data related to a patient's specimen test and patient state data indicating the state of the patient; a specimen test data statistical unit that obtains statistics of the specimen test data; a patient state data classification unit that classifies the patient state data into any one of a plurality of preset categories; a prediction unit that inputs the statistics obtained from the specimen test data and the classification result of the patient state data into a prediction model that is machine learning the relationship between the statistics of the specimen test data, the category of the patient's state, and the infectious disease, and predicts the possibility that the patient is infected with an infectious disease based on the output result from the prediction model; and is characterized by comprising the above.

[0009] Also, to achieve the above object, an infection prediction method according to one aspect of the present invention includes: a data acquisition step of acquiring specimen test data related to a patient's specimen test and patient state data indicating the state of the patient; a specimen test data statistical step of obtaining statistics of the specimen test data; a patient state data classification step of classifying the patient state data into any one of a plurality of preset categories; Input the statistics obtained from the specimen test data and the classification result of the patient status data into a prediction model that is machine learning the relationship between the statistics of the specimen test data, the category of the patient's status, and the infectious disease, and predict the possibility that the patient is infected with the infectious disease based on the output result from the prediction model, a prediction step; It is characterized by having.

[0010] Furthermore, in order to achieve the above object, the program in one aspect of the present invention is To a computer, A data acquisition step of acquiring specimen test data related to the patient's specimen test and patient status data indicating the status of the patient; A specimen test data statisticalization step of obtaining the statistics of the specimen test data; A patient status data classification step of classifying the patient status data into any one of a plurality of preset categories; Input the statistics obtained from the specimen test data and the classification result of the patient status data into a prediction model that is machine learning the relationship between the statistics of the specimen test data, the category of the patient's status, and the infectious disease, and predict the possibility that the patient is infected with the infectious disease based on the output result from the prediction model, a prediction step; It is characterized by causing to execute.

Effect of the Invention

[0011] As described above, according to the present invention, it is possible to predict the possibility of nosocomial infection of a patient, including the cause of the infection.

Brief Description of the Drawings

[0012]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Figure 16

Figure 17

Figure 18

Figure 19

Figure 20

Embodiments for Carrying Out the Invention

[0013] (Embodiment 1) Hereinafter, the infection prediction apparatus, the infection prediction method, and the program according to Embodiment 1 will be described with reference to FIGS. 1 to 8.

[0014] [Device Configuration] First, the schematic configuration of the infection prediction apparatus 100 according to Embodiment 1 will be described with reference to FIG. 1. FIG. 1 is a configuration diagram showing the schematic configuration of the infection prediction apparatus according to Embodiment 1.

[0015] The infection prediction apparatus 100 according to Embodiment 1 shown in FIG. 1 is an apparatus for predicting the possibility of being infected with an infectious disease. As shown in FIG. 1, the infection prediction apparatus 100 includes a data acquisition unit 10, a specimen test data statistical unit 21, a patient state data classification unit 22, and a prediction unit 30.

[0016] The data acquisition unit 10 acquires specimen test data related to the specimen test of the patient and patient state data indicating the state of the patient. The specimen test data statistical unit 21 obtains the statistics of the specimen test data. The patient state data classification unit 22 classifies the patient state data into any one of a plurality of preset categories.

[0017] The prediction unit 30 inputs the statistics obtained from the specimen test data and the classification result of the patient state data into a prediction model that has learned the relationship between the statistics of the specimen test data, the category of the patient state, and the infectious disease. Then, the prediction unit 14 predicts the possibility that the patient is infected with an infectious disease based on the output result from the prediction model.

[0018] As described above, in Embodiment 1, the prediction model used for predicting the possibility of infection learns the statistics of specimen test data and the relationship between the categories of patient conditions and infectious diseases. Therefore, according to Embodiment 1, it is possible to predict the possibility of a patient being infected with an infectious disease, including the cause of the infection.

[0019] Subsequently, with reference to FIGS. 2 to 5, the configuration and functions of the infection prediction apparatus in Embodiment 1 will be specifically described. FIG. 2 is a configuration diagram specifically showing the configuration of the infection prediction apparatus in Embodiment 1. FIG. 3 is a diagram showing an example of the statistical processing of specimen test data related to a patient's specimen test. FIG. 4 is a diagram showing an example of the classification of patient state data indicating the state of a patient. FIG. 5 is a diagram showing an example of the statistical processing of treatment data related to the treatment performed on a patient.

[0020] As shown in FIG. 2, in Embodiment 1, the infection prediction apparatus 100 includes a treatment data statistical unit 23 and a storage unit 40 in addition to the above-described data acquisition unit 10, specimen test data statistical unit 21, patient state data classification unit 22, and prediction unit 30. Among these, the specimen test data statistical unit 21, patient state data classification unit 22, and treatment data statistical unit 23 function as a preprocessing unit 20 that performs preprocessing of data to be input to the prediction model.

[0021] Also, as shown in FIG. 2, in Embodiment 1, as the prediction models, a catheter bloodstream infection prediction model 41, a urinary tract infection prediction model 42, and a causative bacterium prediction model 43 are used, and these are stored in the storage unit 40.

[0022] As shown in FIG. 2, the data acquisition unit 10 further acquires treatment data related to the treatment performed on the patient, in addition to the above-described specimen test data and patient state data. Examples of the data acquisition sources of the data acquisition unit 10 include a terminal device that is data communicably connected by a network or the like, and an input device such as a keyboard.

[0023] As shown in FIG. 3, examples of the specimen examination data include data obtained from blood tests such as a patient's white blood cells, red blood cells, and platelets. In the example of FIG. 3, the data obtained from the white blood cell count examination conducted during period 1 from the patient A's hospitalization to catheter insertion and period 2 from the patient A's catheter insertion until the suspicion of an infectious disease is shown as the specimen examination data.

[0024] In Embodiment 1, as shown in FIG. 3 for example, the specimen examination data statistical unit 21 specifies the maximum value and the minimum value from the white blood cell count examination results for each period, and further calculates the average value, and uses the obtained maximum value, minimum value, and average value as the statistics of the specimen examination data. Note that the specimen examination data statistical unit 21 may use only any one of the maximum value, minimum value, and average value as the statistic, or variance or the like may be used instead of the average value.

[0025] As shown in FIG. 4, examples of the patient state data include, for example, the patient's body temperature, blood pressure (systolic), pulse (HR: Heart Rate), oxygen saturation (SPO2), etc. on the day when an infectious disease is suspected. In addition, the patient state data may include items other than these, or any of these may be omitted. Furthermore, the patient state data may include the patient's personal information, for example, age, gender, hospitalization period, presence or absence of catheter insertion, etc.

[0026] In Embodiment 1, as shown in FIG. 4 for example, the patient state data classification unit 22 classifies each value of the patient state data into any of the categories from “-3” to “3” according to the preset classification rules. Note that the categories may be set based on, for example, NEWS (National Early Warning Score) shown in the following reference document. Also, when the above-mentioned patient's personal information is included in the patient state data, the personal information may not be classified. Reference document: https: / / www.jstage.jst.go.jp / article / jsicm / 25 / 6 / 25_25_453 / _pdf / -char / en

[0027] As shown in FIG. 5, examples of the treatment data include data indicating, for example, the implementation date of dental treatment performed on a patient. In the example of FIG. 5, the dental treatment performed during period 1 from the hospitalization of patient A to catheter insertion and period 2 from catheter insertion of patient A to the occurrence of suspicion of infectious disease is shown as the treatment data.

[0028] The treatment data statistical unit 23 calculates the statistics of the treatment data acquired by the data acquisition unit 10. In the first embodiment, the treatment data statistical unit 23, for example, as shown in FIG. 5, calculates the number of dental treatments performed in each period, and uses the calculated number of dental treatments for each period as the statistics. Note that in the first embodiment, the treatment targeted by the treatment data is not limited to dental treatment.

[0029] In the first embodiment, the prediction model performs machine learning on the relationship between the statistics of the specimen test data, the category of the patient's condition, the statistics of the treatment data, and the infectious disease. As shown in FIG. 2, as the prediction models, a catheter bloodstream infection prediction model 41, a urinary tract infection prediction model 42, and a causative bacterium prediction model 43 are used.

[0030] The catheter bloodstream infection prediction model 41 is a machine learning model constructed by performing machine learning on the relationship between the statistics of the specimen test data, the category of the patient's condition, the statistics of the treatment data, and the catheter infection.

[0031] The urinary tract infection prediction model 42 is a machine learning model constructed by performing machine learning on the relationship between the statistics of the specimen test data, the category of the patient's condition, and the urinary tract infection.

[0032] The causative bacterium prediction model 43 is a machine learning model constructed by performing machine learning on the relationship between the statistics of the specimen test data, the category of the patient's condition, the statistics of the treatment data, and the causative bacterium of the catheter bloodstream infection.

[0033] In Embodiment 1, the prediction unit 30 inputs statistics obtained from specimen test data, classification results of patient state data, and statistics obtained from treatment data into prediction models such as a catheter bloodstream infection prediction model 41, a urinary tract infection prediction model 42, and a causative bacterium prediction model 43. Then, based on the output results from each prediction model, the prediction unit 30 predicts the possibility that the patient is infected with an infectious disease. The processing in the prediction unit 30 will be described later.

[0034] Here, with reference to FIG. 6, machine learning in the catheter bloodstream infection prediction model 41, the urinary tract infection prediction model 42, and the causative bacterium prediction model 43 will be described. FIG. 6 is a diagram showing an example of training data used for machine learning in Embodiment 1.

[0035] As shown in FIG. 6, the training data for the catheter bloodstream infection prediction model 41 has the presence or absence of catheter bloodstream infection as the correct label and is composed of the age, gender, length of hospitalization, white blood cell count, blood pressure category, and number of dental treatments of the patients at the data acquisition source.

[0036] Also, as shown in FIG. 6, the training data for the urinary tract infection prediction model 42 has the presence or absence of urinary tract infection as the correct label and is composed of the age, gender, length of hospitalization, C-reactive protein (CRP), and oxygen saturation category of the patients at the data acquisition source.

[0037] Furthermore, as shown in FIG. 6, the training data for the causative bacterium prediction model 43 has whether or not the causative agent is CNS (coagulase-negative bacteria) as the correct label and is composed of the age, gender, length of hospitalization, white blood cell count, blood pressure category, and number of dental treatments of the patients at the data acquisition source.

[0038] Machine learning for each prediction model is performed, for example, by inputting the explanatory variables of the training data into a convolutional neural network, a graph convolutional network, a support vector machine, a regression model, etc., and updating the parameters so that the output value matches the correct label.

[0039] [Operation of the device] Next, the operation of the infection prediction device 100 in Embodiment 1 will be described with reference to FIG. 7. FIG. 7 is a flowchart showing the operation of the infection prediction device in Embodiment 1. In the following description, FIGS. 1 to 6 will be referred to as appropriate. In Embodiment 1, by operating the infection prediction device 100, an infection prediction method is implemented. Therefore, the description of the infection prediction method in Embodiment 1 will be replaced with the following description of the operation of the infection prediction device 100.

[0040] First, as shown in FIG. 7, the data acquisition unit 10 acquires specimen test data, patient status data, and treatment data (step A1). Then, the data acquisition unit 10 inputs the acquired specimen test data, patient status data, and treatment data to the preprocessing unit 20.

[0041] Next, the specimen test data statistical unit 21 calculates the statistics of the specimen test data acquired in step A1 (step A2). Specifically, as shown in FIG. 3, for example, the specimen test data statistical unit 21 identifies the maximum value and the minimum value from the test results of the white blood cell count for each period, and further calculates the average value.

[0042] Next, the patient status data classification unit 22 classifies the patient status data of the patient acquired in step A1 into any one of a plurality of preset categories (step A3). Specifically, as shown in FIG. 4, for example, the patient status data classification unit 22 classifies the patient's body temperature, blood pressure (systolic), pulse (HR: Heart Rate), oxygen saturation, etc. into any of the categories from "-3" to "3" according to the preset classification rules.

[0043] Next, the treatment data statistical unit 23 calculates the statistics of the treatment data acquired in step A1 (step A4). Specifically, as shown in FIG. 5, for example, the treatment data statistical unit 23 obtains the number of dental treatments performed in each period, and uses the obtained number of dental treatments per period as the statistic.

[0044] Next, the prediction unit 30 inputs the statistics obtained in steps A2 to A4 into the catheter bloodstream infection prediction model 41, the urinary tract infection prediction model 42, and the causative bacterium prediction model 43. Then, based on the output results from each prediction model, the prediction unit 30 predicts the possibility that the patient is infected with an infectious disease and outputs the prediction result (step A5).

[0045] Subsequently, with reference to FIG. 8, the prediction process (step A5) shown in FIG. 7 will be described in more detail. FIG. 8 is a flowchart specifically showing the prediction process shown in FIG. 7.

[0046] As shown in FIG. 8, first, the prediction unit 30 determines whether a catheter has been inserted into the patient (step A51). If notification of the presence or absence of catheter insertion is received from the outside, the prediction unit 30 makes a determination based on the notification. If the personal information of the patient included in the patient status data contains information on the presence or absence of catheter insertion, the prediction unit 30 makes a determination based on the patient status data.

[0047] As a result of the determination in step A51, if a catheter has been inserted into the patient (step A51: Yes), the prediction unit 30 inputs the statistics obtained in step A2, the statistics obtained in step A3 (classification result), and the statistics obtained in step A4 into the catheter bloodstream infection prediction model 41 (step A52).

[0048] Since the catheter bloodstream infection prediction model 41 outputs "catheter bloodstream infection" or "non-infectious disease" as the output, the prediction unit 30 determines whether it is positive based on the output result of the catheter bloodstream infection prediction model 41 (step A53).

[0049] As a result of the determination in step A53, if it is not positive (step A53: No), the prediction unit 30 determines that there is no possibility of an infectious disease and outputs the determination result as the prediction result (step A56).

[0050] On the other hand, if the result of the determination in step A53 is positive (step A53: Yes), the prediction unit 30 inputs the statistics obtained in step A2, the statistics obtained in step A3, and the statistics obtained in step A4 into the causative bacterium prediction model 43, and executes the prediction (step A54).

[0051] Since the causative bacterium prediction model 43 outputs whether or not the cause is CNS according to the input, after executing step A54, the prediction unit 30 outputs the output result of "CNS is the cause" or "CNS is not the cause" as the prediction result (step A56).

[0052] Also, if the result of the determination in step A51 described above is that the catheter has not been inserted into the patient (step A51: No), the prediction unit 30 inputs the statistics obtained in step A2 and the classification result obtained in step A3 into the urinary tract infection prediction model 42 (step A55).

[0053] Then, since the urinary tract infection prediction model 42 outputs "urinary tract infection" or "non-infection" as the output, the prediction unit 30 determines the possibility of urinary tract infection based on the output result of the urinary tract infection prediction model 42. The prediction unit 30 outputs the output result of the urinary tract infection prediction model 42 as the prediction result (step A56).

[0054] As described above, in the first embodiment, the catheter bloodstream infection prediction model 41, the urinary tract infection prediction model 42, and the causative bacterium prediction model 43 are used. Therefore, according to the first embodiment, it is possible to predict the possibility of infection of the patient including the cause of the infection.

[0055] [Program] The program in Embodiment 1 may be any program that causes a computer to execute Steps A1 to A5 shown in FIG. 7. By installing and executing this program on a computer, the infection prediction device 100 and the infection prediction method in Embodiment 1 can be realized. In this case, the processor of the computer functions as the data acquisition unit 10, the preprocessing unit 20, and the prediction unit 30, and performs processing.

[0056] Also, in Embodiment 1, the storage unit 40 may be realized by storing data files constituting these in a storage device such as a hard disk provided in the computer, or may be realized by a storage device of another computer. Examples of the computer include general-purpose PCs, smartphones, and tablet terminal devices.

[0057] The program in Embodiment 1 may be executed by a computer system constructed by a plurality of computers. In this case, for example, each computer may function as any one of the data acquisition unit 10, the preprocessing unit 20, and the prediction unit 30.

[0058] (Embodiment 2) Next, the infection prediction device, the infection prediction method, and the program in Embodiment 2 will be described with reference to FIGS. 9 to 13.

[0059] [Device Configuration] First, the configuration of the infection prediction device 200 in Embodiment 2 will be described with reference to FIG. 9. Also, the functions of the infection prediction device 200 in Embodiment 2 will be described with reference to FIGS. 10 and 11. FIG. 9 is a configuration diagram specifically showing the configuration of the infection prediction device in Embodiment 2. FIG. 10 is a diagram showing an example of the statisticalization of drug data regarding the prescription of drugs given to a patient. FIG. 11 is a diagram showing an example of the statisticalization of physical condition data regarding the physical condition of a patient.

[0060] The infection prediction device 200 in Embodiment 2 shown in FIG. 9 includes a data acquisition unit 10, a preprocessing unit 20, a prediction unit 30, and a storage unit 40, similar to the infection prediction device 100 in Embodiment 1. However, the infection prediction device 200 in Embodiment 2 is different from the infection prediction device 100 in Embodiment 1 in the configuration of the preprocessing unit 20. Hereinafter, the description will focus on the differences from Embodiment 1.

[0061] As shown in FIG. 9, in Embodiment 2, the infection prediction device 200, in addition to the specimen examination data statistical unit 21, the patient status data classification unit 22, and the treatment data statistical unit 23 described in Embodiment 1, in the preprocessing unit 20, further includes a drug data statistical unit 24 and a physical condition data statistical unit 25.

[0062] In Embodiment 2, the data acquisition unit 10 acquires, in addition to the specimen examination data, patient status data, and treatment data described above, drug data regarding the prescription of drugs administered to the patient and physical condition data regarding the patient's physical condition.

[0063] As the drug data, for example, data indicating what drugs were prescribed to the patient and when is used. The upper part of FIG. 10 shows an example of drug data.

[0064] The drug data statistical unit 24 calculates the statistics of the drug data acquired by the data acquisition unit 10. For example, as shown in the middle part of FIG. 10, the drug data statistical unit 24 first sets, as the periods, period 1 from the patient A's hospitalization to catheter insertion and period 2 from the patient A's catheter insertion to the occurrence of suspicion of infection, and measures the number of times of drug prescriptions performed in each period.

[0065] Subsequently, as shown in the lower part of FIG. 10, the drug data statistical unit 24 divides the number of times of drug prescriptions by the number of days in each period for each period, and uses the obtained value for each period as the statistics of the drug data. For example, if drug A was prescribed 5 times in a certain period and the number of days in that period is 10 days, the statistic is 0.5.

[0066] In addition, the statistics of the drug data shown in FIG. 10 use the catheter insertion time as the period delimiter and are used as the input to the catheter bloodstream infection prediction model 41. However, in the second embodiment, the statistics of the drug data can also be used as the input to the urinary tract infection prediction model 42. In this case, the drug data statistical unit 24 sets the period from the patient's hospitalization until the suspicion of infection occurs as period 1, and the period from the occurrence of the suspicion of infection until discharge as period 2.

[0067] As the physical condition data, for example, as shown in the upper part of FIG. 11, data indicating the daily stool volume of the patient is used. In the example of FIG. 11, the stool volume is evaluated in 10 levels from 0 to 10 according to the amount, for example.

[0068] First, the physical condition data statistical unit 25 unifies "1" to "10" as "1" and converts the daily stool volume into binary data of "0" and "1", as shown in the middle part of FIG. 11, for example. Further, as shown in the lower part of FIG. 11, the physical condition data statistical unit 25 determines whether "0" exceeds 50% during the period from the patient's hospitalization until the suspicion of infection occurs. Then, if it exceeds, the physical condition data during that period is set as "0: no stool", and if it does not exceed, the physical condition data during that period is set as "1: having stool". The obtained result becomes the statistics of the physical condition data. Also, in the second embodiment, the statistics of the physical condition data are used as the input to the urinary tract infection prediction model 42.

[0069] Here, with reference to FIG. 12, the machine learning in the catheter bloodstream infection prediction model 41, the urinary tract infection prediction model 42, and the causative bacterium prediction model 43 in the second embodiment will be described. FIG. 12 is a diagram showing an example of the training data used for machine learning in the second embodiment.

[0070] As shown in Fig. 12, the training data for the catheter bloodstream infection prediction model 41 uses the presence or absence of catheter bloodstream infection as the correct label, similar to the example shown in Fig. 6. However, in the second embodiment, different from the example shown in Fig. 6, the training data for the catheter bloodstream infection prediction model 41 includes, in addition to the age, gender, length of hospitalization, white blood cell count, blood pressure category, and number of dental treatments of the patients from whom the data is obtained, the number of surgeries and the average number of prescriptions per period of the drug as components.

[0071] Also, as shown in Fig. 12, the training data for the urinary tract infection prediction model 42 uses the presence or absence of urinary tract infection as the correct label, similar to the example shown in Fig. 6. However, in the second embodiment, different from the example shown in Fig. 6, the training data for the urinary tract infection prediction model 42 includes, in addition to the age, gender, length of hospitalization, C-reactive protein (CRP), and oxygen saturation category of the patients from whom the data is obtained, the presence or absence of feces as a component.

[0072] Furthermore, in the second embodiment, when the statistics of drug data are used as an input to the urinary tract infection prediction model 42, the training data for the urinary tract infection prediction model 42 may include the average number of prescriptions per period of the drug as a component.

[0073] Note that, as shown in Fig. 12, also in the second embodiment, the training data for the causative bacterium prediction model 43 is the same as the example shown in Fig. 6. The training data for the causative bacterium prediction model 43 uses whether or not the causative agent is CNS (coagulase-negative bacteria) as the correct label and is composed of the age, gender, length of hospitalization, white blood cell count, blood pressure category, and number of dental treatments of the patients from whom the data is obtained.

[0074] The machine learning of each prediction model is performed in the same manner as in the first embodiment. For example, machine learning is performed by inputting the explanatory variables of the training data into a convolutional neural network, a graph convolutional network, a support vector machine, a regression model, etc., and updating the parameters so that the output value matches the correct label.

[0075] [Device Operation] Next, the operation of the infection prediction device 200 in Embodiment 2 will be described with reference to FIG. 13. FIG. 13 is a flowchart showing the operation of the infection prediction device in Embodiment 2. In the following description, FIGS. 9 to 12 will be referred to as appropriate. In Embodiment 2, an infection prediction method is implemented by operating the infection prediction device 200. Therefore, the description of the infection prediction method in Embodiment 2 will be replaced with the following description of the operation of the infection prediction device 200.

[0076] First, as shown in FIG. 13, the data acquisition unit 10 acquires specimen test data, patient status data, treatment data, drug data, and physical condition data (step B1). Then, the data acquisition unit 10 inputs the acquired various data to the preprocessing unit 20.

[0077] Next, the specimen test data statistical unit 21 obtains the statistics of the specimen test data acquired in step B1 (step B2). Step B2 is the same as step A2 shown in FIG. 7.

[0078] Next, the patient status data classification unit 22 classifies the patient status data of the patient acquired in step B1 into any one of a plurality of preset categories (step B3). Step B3 is the same as step A3 shown in FIG. 7.

[0079] Next, the treatment data statistical unit 23 obtains the statistics of the treatment data acquired in step B1 (step B4). Step B4 is the same as step A4 shown in FIG. 7.

[0080] Next, the drug data statistical unit 24 obtains the statistics of the drug data acquired in step B1 (step B5).

[0081] Specifically, the drug data statistical unit 24 first sets a period, measures the number of times of drug prescriptions performed in each period, and then, for each period, divides the measured number of times by the number of days in each period, and sets the obtained value for each period as the statistics of the drug data.

[0082] Next, the physical condition data statistical unit 25 calculates the statistics of the physical condition data acquired in step B1 (step B6).

[0083] Specifically, the physical condition data statistical unit 25 first converts the physical condition data in which the amount of feces is indicated in 10 levels into binary data in which the amount of feces of a patient in a day is expressed as "0" and "1". Subsequently, the physical condition data statistical unit 25 determines whether "0" exceeds 50% during the set period. If it exceeds, the physical condition data during that period is set to "0: no feces", and if it does not exceed, the physical condition data during that period is set to "1: having feces" and statistically processed.

[0084] Next, the prediction unit 30 inputs the statistics obtained in steps B2 to B6 into the catheter bloodstream infection prediction model 41, the urinary tract infection prediction model 42, and the causative bacterium prediction model 43. Then, the prediction unit 30 predicts the possibility that the patient is infected with an infectious disease based on the output results from each prediction model and outputs the prediction result (step B7).

[0085] The prediction process in step B7 is performed according to the steps shown in FIG. 8 in Embodiment 1. However, in Embodiment 2, in step A53, the prediction unit 30 inputs, into the catheter bloodstream infection prediction model 41, the statistics obtained in steps B2 to B4 and also the statistics of the drug data obtained in step B5.

[0086] Furthermore, in Embodiment 2, in step A55, the prediction unit 30 inputs, into the urinary tract infection prediction model 42, the statistics obtained in steps B2 and B3 and also the statistics of the physical condition data obtained in step B6.

[0087] As described above, in Embodiment 2, different from Embodiment 1, drug data and physical condition data are also used, and these statistics are used for prediction using the prediction model. Therefore, according to Embodiment 2, it is possible to further predict the possibility of the patient being infected.

[0088] [Program] The program in Embodiment 2 may be any program that causes a computer to execute Steps B1 to B7 shown in FIG. 13. By installing and executing this program on a computer, the infection prediction device 200 and the infection prediction method in Embodiment 2 can be realized. In this case, the processor of the computer functions as the data acquisition unit 10, the preprocessing unit 20, and the prediction unit 30, and performs processing.

[0089] Also, in Embodiment 2, the storage unit 40 may be realized by storing the data files constituting these in a storage device such as a hard disk provided in the computer, or may be realized by a storage device of another computer. Examples of the computer include a general-purpose PC, a smartphone, and a tablet terminal device.

[0090] The program in Embodiment 2 may also be executed by a computer system constructed by a plurality of computers. In this case, for example, each computer may function as any one of the data acquisition unit 10, the preprocessing unit 20, and the prediction unit 30.

[0091] (Embodiment 3) Next, the infection prediction device, the infection prediction method, and the program in Embodiment 3 will be described with reference to FIGS. 14 to 19.

[0092] [Device Configuration] First, the configuration of the infection prediction device 300 in Embodiment 3 will be described with reference to FIG. 14. Also, the functions of the infection prediction device 300 in Embodiment 3 will be described with reference to FIGS. 15 and 16. FIG. 14 is a configuration diagram specifically showing the configuration of the infection prediction device in Embodiment 3. FIG. 15 is a diagram showing an example of the statistical processing of surgical data regarding the surgery performed on a patient. FIG. 16 is a diagram showing an example of the statistical processing of disease name data regarding the disease name registered for a patient.

[0093] As shown in FIG. 14, the infection prediction device 300 in Embodiment 3 includes a data acquisition unit 10, a preprocessing unit 20, a prediction unit 30, and a storage unit 40, similar to the infection prediction devices in Embodiments 1 and 2. However, the infection prediction device 300 in Embodiment 3 differs from the infection prediction devices in Embodiments 1 and 2 in the configuration of the preprocessing unit 20. Hereinafter, the description will focus on the differences from Embodiments 1 and 2.

[0094] As shown in FIG. 14, in Embodiment 3, the infection prediction device 300 includes, in the preprocessing unit 20, in addition to the specimen examination data statistical unit 21, patient status data classification unit 22, treatment data statistical unit 23, drug data statistical unit 24, and physical condition data statistical unit 25 described in Embodiment 2, a surgical data statistical unit 26 and a disease name data statistical unit 27.

[0095] In Embodiment 3, the data acquisition unit 10 acquires, in addition to the specimen examination data, patient status data, treatment data, drug data, and physical condition data described above, surgical data related to the surgery performed on the patient and disease name data related to the disease name registered for the patient.

[0096] As the surgical data, for example, data indicating when the surgery was performed on the patient is used. The upper part of FIG. 15 shows an example of the surgical data.

[0097] The surgical data statistical unit 26 obtains the statistics of the surgical data acquired by the data acquisition unit 10. For example, as shown in the lower part of FIG. 15, the surgical data statistical unit 26 first sets, as the period, period 1 from the patient A's hospitalization to catheter insertion and period 2 from the patient A's catheter insertion to the occurrence of suspicion of an infectious disease, and measures the number of surgeries performed in each period. Then, the surgical data statistical unit 26 stores the measured number of surgeries per period in the storage unit 40 as the statistical surgical data.

[0098] As disease name data, for example, data indicating what disease names were registered for a patient at what time is used. The upper part of FIG. 16 designates an example of disease name data.

[0099] The disease name data statistical unit 27 calculates the statistics of the disease name data acquired by the data acquisition unit 10. For example, as shown in the middle part of FIG. 16, the disease name data statistical unit 27 first sets, as periods, period 1 from the hospitalization of patient A to the catheter insertion, and period 2 from the catheter insertion of patient A to the time when suspicion of an infectious disease occurred.

[0100] Subsequently, as shown in the middle part of FIG. 16, the disease name data statistical unit 27 converts the registered disease names into codes using a classification table prepared in advance. Examples of the classification table used in this case include the ICD (International Statistical Classification of Diseases and Related Health Problems) international classification of diseases (www.byomei.org / icd10 / index.html).

[0101] Next, as shown in the lower part of FIG. 16, the disease name data statistical unit 27 measures the number of times a specific code was registered for each period using the code-converted disease name data. The number of registrations for each period becomes the statistics of the disease name data. Also, the statistics of the disease name data are used as input to the SSI (Surgical Site Infection) prediction model 44.

[0102] Here, the SSI prediction model 44 will be described with reference to FIG. 17. FIG. 17 is a diagram showing an example of training data used for machine learning of the SSI prediction model in Embodiment 3. In Embodiment 3, as the catheter bloodstream infection prediction model 41, the urinary tract infection prediction model 42, and the causative bacterium prediction model 43, the prediction models shown in Embodiment 3 are used.

[0103] The SSI prediction model 44 is a machine learning model constructed by machine learning the relationships between the statistics of specimen examination data, the categories of patient conditions, the statistics of treatment data, the statistics of disease name data, and the causative bacteria of postoperative infectious diseases.

[0104] As shown in FIG. 17, the training data for the SSI prediction model 44 uses whether or not it corresponds to SSI (postoperative infectious disease) as the correct label, and is composed of the age, gender, length of hospital stay, white blood cell count, body temperature, number of surgeries, average number of prescription times of drugs during a period, and the number of registrations of disease names during a specific period of the patient from whom the data is acquired. Also, as the correct label, a probability value (%) indicating the possibility of being SSI may be used.

[0105] The machine learning of the SSI prediction model 44 is also performed in the same manner as in Embodiment 1. For example, the explanatory variables of the training data are input into a convolutional neural network, a graph convolutional network, a support vector machine, a regression model, etc., and the machine learning is performed by updating the parameters so that the output value matches the correct label.

[0106] [Device Operation] Next, the operation of the infection prediction device 300 in Embodiment 3 will be described with reference to FIG. 18. FIG. 18 is a flowchart showing the operation of the infection prediction device in Embodiment 3. In the following description, FIGS. 14 to 17 will be referred to as appropriate. In Embodiment 3, the infection prediction method is implemented by operating the infection prediction device 300. Therefore, the description of the infection prediction method in Embodiment 3 will be replaced with the following description of the operation of the infection prediction device 300.

[0107] First, as shown in FIG. 18, the data acquisition unit 10 acquires specimen examination data, patient state data, treatment data, drug data, physical condition data, surgical data, and disease name data (step C1). Then, the data acquisition unit 10 inputs the acquired various data into the preprocessing unit 20.

[0108] Next, the specimen examination data statistical unit 21 calculates the statistics of the specimen examination data obtained in step C1 (step C2). Step C2 is the same as step A2 shown in FIG. 7.

[0109] Next, the patient status data classification unit 22 classifies the patient status data of the patient obtained in step C1 into any one of a plurality of preset categories (step C3). Step C3 is the same as step A3 shown in FIG. 7.

[0110] Next, the treatment data statistical unit 23 calculates the statistics of the treatment data obtained in step C1 (step C4). Step C4 is the same as step A4 shown in FIG. 7.

[0111] Next, the drug data statistical unit 24 calculates the statistics of the drug data obtained in step C1 (step C5). Step C5 is the same as step B5 shown in FIG. 13.

[0112] Next, the physical condition data statistical unit 25 calculates the statistics of the physical condition data obtained in step C1 (step C6). Step C6 is the same as step B6 shown in FIG. 13.

[0113] Next, the surgical data statistical unit 26 calculates the statistics of the surgical data obtained in step C1 (step C7).

[0114] Specifically, the surgical data statistical unit 26 first sets a period, measures the number of surgeries performed in each period, and then stores the number of surgeries performed per period in the storage unit 40 as the statistical surgical data.

[0115] Next, the disease name data statistical unit 27 calculates the statistics of the disease name data obtained in step C1 (step C8).

[0116] Specifically, the disease name data statistical unit 27 first sets a period and converts the registered disease names into codes using a pre-prepared classification table. Subsequently, the disease name data statistical unit 27 measures the number of times a specific code is registered for each period using the code-converted disease name data. The registration count for each period becomes the statistics of the disease name data.

[0117] Next, the prediction unit 30 inputs the statistics obtained in steps C2 to C6 and C8 into the catheter bloodstream infection prediction model 41, the urinary tract infection prediction model 42, the causative bacteria prediction model 43, and the SSI prediction model 44. Then, based on the output results from each prediction model, the prediction unit 30 predicts the possibility that the patient is infected with an infectious disease and outputs the prediction result (step C9).

[0118] Subsequently, with reference to FIG. 19, the prediction process (step C9) shown in FIG. 18 will be described in more detail. FIG. 19 is a flowchart specifically showing the prediction process shown in FIG. 18.

[0119] As shown in FIG. 19, first, the prediction unit 30 determines whether a catheter has been inserted into the patient (step C101). If notification of the presence or absence of catheter insertion is received from the outside, the prediction unit 30 makes a determination based on the notification. If the personal information of the patient added to the patient status data includes the presence or absence of catheter insertion, the prediction unit 30 makes a determination based on the patient status data.

[0120] As a result of the determination in step C101, if a catheter has been inserted into the patient (step C101: Yes), the prediction unit 30 inputs the statistics obtained in step C2, the classification result obtained in step C3, the statistics obtained in step C4, and the statistics of the drug data obtained in step C5 into the catheter bloodstream infection prediction model 41 (step C102).

[0121] Then, since the catheter bloodstream infection prediction model 41 outputs "catheter bloodstream infection" or "non-infection" as an output, the prediction unit 30 determines whether it is positive based on the output result of the catheter bloodstream infection prediction model 41 (step C103).

[0122] If the result of the determination in step C103 is not positive (step C103: No), the prediction unit 30 determines that there is no possibility of infection and outputs the determination result as a prediction result (step C108).

[0123] On the other hand, if the result of the determination in step C103 is positive (step C103: Yes), the prediction unit 30 inputs the statistics obtained in step C, the classification result obtained in step C3, and the statistics obtained in step C4 into the causative bacterium prediction model 43 and executes the prediction (step C104).

[0124] Since the causative bacterium prediction model 43 outputs whether CNS is the cause according to the input, after executing step C104, the prediction unit 30 outputs the output result of "CNS is the cause" or "CNS is not the cause" as a prediction result (step C108).

[0125] Also, if the result of the determination in step C101 described above is that the catheter has not been inserted into the patient (step C101: No), the prediction unit 30 uses the statistical operation data stored in the storage unit 40 and the disease name data obtained in step C1 to determine whether the patient has undergone surgery and whether inflammation has occurred in the patient (step C105).

[0126] If the result of the determination in step C105 corresponds to "the patient has undergone surgery and inflammation has occurred in the patient" (step C105: Yes), the prediction unit 30 inputs the statistics obtained in step C2, the classification result obtained in step C3, the statistics obtained in step C4, and the statistics of the disease name data obtained in step C8 into the SSI prediction model 44 (step C106).

[0127] Since the SSI prediction model 44 outputs whether it corresponds to SSI according to the input, after executing step C106, the prediction unit 30 outputs the output result of "the patient has SSI (postoperative infectious disease)" or "the patient has no infectious disease" as the prediction result (step C108).

[0128] On the other hand, if the result of the determination in step C105 does not correspond to "surgery is performed on the patient and inflammation has occurred in the patient" (step C105: No), the prediction unit 30 inputs the statistics obtained in step C2, the classification result obtained in step C3, and the statistics of the physical condition data obtained in step C6 into the urinary tract infection prediction model 42 (step C107).

[0129] Since the urinary tract infection prediction model 42 outputs "urinary tract infection" or "no infectious disease" as the output, the prediction unit 30 determines the possibility of urinary tract infection based on the output result of the urinary tract infection prediction model 42. The prediction unit 30 outputs the output result of the urinary tract infection prediction model 42 as the prediction result (step C108).

[0130] As described above, in the third embodiment, the catheter bloodstream infection prediction model 41, the urinary tract infection prediction model 42, the causative bacterium prediction model 43, and the SSI prediction model 44 are used. Therefore, according to the third embodiment, it is possible to predict the possibility of a patient's infection including the cause of postoperative infection.

[0131] As described above, in the third embodiment, different from the first and second embodiments, surgical data and disease name data are used to predict whether the patient is infected with a postoperative infectious disease. Therefore, according to the third embodiment, it is possible to further predict the possibility of a patient's infection.

[0132] [Program] The program in Embodiment 3 may be any program that causes a computer to execute Steps C1 to C9 shown in FIG. 18. By installing and executing this program on a computer, the infection prediction device 300 and the infection prediction method in Embodiment 3 can be realized. In this case, the processor of the computer functions as the data acquisition unit 10, the preprocessing unit 20, and the prediction unit 30, and performs processing.

[0133] Also, in Embodiment 3, the storage unit 40 may be realized by storing data files constituting these in a storage device such as a hard disk provided in the computer, or may be realized by a storage device of another computer. Examples of the computer include general-purpose PCs, smartphones, and tablet terminal devices.

[0134] The program in Embodiment 3 may also be executed by a computer system constructed by a plurality of computers. In this case, for example, each computer may function as any one of the data acquisition unit 10, the preprocessing unit 20, and the prediction unit 30.

[0135] [Physical Configuration] Here, a computer that realizes an infection prediction device by executing the programs in Embodiments 1 to 3 will be described with reference to FIG. 20. FIG. 20 is a block diagram showing an example of a computer that realizes the infection prediction device in Embodiments 1 to 3.

[0136] As shown in FIG. 20, the computer 110 includes a CPU (Central Processing Unit) 111, a main memory 112, a storage device 113, an input interface 114, a display controller 115, a data reader / writer 116, and a communication interface 117. These units are connected to each other via a bus 121 so as to be able to communicate data with each other.

[0137] In addition to, or instead of, the CPU 111, the computer 110 may include a GPU (Graphics Processing Unit) or an FPGA (Field-Programmable Gate Array). In this aspect, the GPU or FPGA can execute the program in the embodiment.

[0138] The CPU 111 expands the program in the embodiment composed of a code group stored in the storage device 113 into the main memory 112 and executes each code in a predetermined order to perform various operations. The main memory 112 is typically a volatile storage device such as a DRAM (Dynamic Random Access Memory).

[0139] Also, the program in the embodiment is provided in a state stored in a computer-readable recording medium 120. Note that the program in this embodiment may circulate on the Internet connected via the communication interface 117.

[0140] Specific examples of the storage device 113 include, in addition to a hard disk drive, a semiconductor storage device such as a flash memory. The input interface 114 mediates data transmission between the CPU 111 and input devices 118 such as a keyboard and a mouse. The display controller 115 is connected to the display device 119 and controls the display on the display device 119.

[0141] The data reader / writer 116 mediates data transmission between the CPU 111 and the recording medium 120, and reads the program from the recording medium 120 and writes the processing result in the computer 110 to the recording medium 120. The communication interface 117 mediates data transmission between the CPU 111 and other computers.

[0142] As specific examples of the recording medium 120, general-purpose semiconductor memory devices such as CF (Compact Flash (registered trademark)) and SD (Secure Digital), magnetic recording media such as Flexible Disk, or optical recording media such as CD-ROM (Compact Disk Read Only Memory) can be mentioned.

[0143] Note that the infection prediction device in Embodiments 1 to 3 can also be realized by using hardware corresponding to each part instead of a computer in which a program is installed. Furthermore, the infection prediction device may be partially realized by a program and the remaining part may be realized by hardware.

[0144] Some or all of the above-described embodiments can be expressed by (Supplementary Note 1) to (Supplementary Note 15) described below, but are not limited to the following description.

[0145] (Supplementary Note 1) A data acquisition unit that acquires specimen test data related to a patient's specimen test and patient state data indicating the state of the patient, A specimen test data statistical unit that obtains statistics of the specimen test data, A patient state data classification unit that classifies the patient state data into any one of a plurality of preset categories, The statistics obtained from the specimen test data and the classification result of the patient state data are input into a prediction model that machine-learns the relationship between the statistics of the specimen test data, the category of the patient's state, and the infectious disease, and based on the output result from the prediction model, the possibility that the patient is infected with the infectious disease is predicted. A prediction unit, An infection prediction device, characterized by comprising:

[0146] (Supplementary Note 2) The infection prediction device according to Supplementary Note 1, wherein the data acquisition unit further acquires treatment data related to the treatment performed on the patient, When the infection prediction device further includes a treatment data statistical unit that calculates the statistics of the treatment data, the prediction model has learned the relationship between the statistics of specimen test data, the categories of the patient's condition, and the statistics of treatment data, and the infectious disease, the prediction unit inputs, into the prediction model, the statistics obtained from the specimen test data and the classification result of the patient state data, as well as the statistics obtained from the treatment data, and predicts the possibility that the patient is infected with an infectious disease based on the output result from the prediction model. An infection prediction device characterized by the above.

[0147] (Appendix 3) The infection prediction device according to Appendix 2, wherein the prediction unit uses, as prediction models, a first prediction model that has learned the relationship between the statistics of specimen test data, the categories of the patient's condition, and the statistics of treatment data, and catheter infectious diseases, and a second prediction model that has learned the relationship between the statistics of specimen test data, the categories of the patient's condition, and the statistics of treatment data, and urinary tract infectious diseases. An infection prediction device characterized by the above.

[0148] (Appendix 4) The infection prediction device according to any one of Appendices 1 to 3, wherein the data acquisition unit acquires the data of the patient's blood test as the specimen test data, and the specimen test data statistical unit calculates, as the statistics of the specimen test data, any one of the maximum value, the minimum value, and the average value. An infection prediction device characterized by the above.

[0149] (Appendix 5) The infection prediction device according to any one of Appendices 1 to 4, wherein the data acquisition unit acquires at least one of the patient's body temperature, blood pressure, pulse, and oxygen saturation as the patient state data, The patient state data classification unit classifies any one of the acquired body temperature, blood pressure, pulse, and oxygen saturation of the patient into any one of a plurality of preset categories according to the value. An infection prediction device characterized by this.

[0150] (Appendix 6) A data acquisition step of acquiring specimen test data related to the patient's specimen test and patient state data indicating the state of the patient; A specimen test data statisticalization step of obtaining the statistics of the specimen test data; A patient state data classification step of classifying the patient state data into any one of a plurality of preset categories; Input the statistics obtained from the specimen test data and the classification result of the patient state data into a prediction model that is machine learning the relationship between the statistics of the specimen test data, the category of the patient's state, and the infectious disease, and predict the possibility that the patient is infected with the infectious disease based on the output result from the prediction model. An infection prediction method characterized by having this.

[0151] (Appendix 7) The infection prediction method according to Appendix 6, In the data acquisition step, further acquire treatment data related to the treatment performed on the patient, The infection prediction method further has a treatment data statisticalization step of obtaining the statistics of the treatment data, The prediction model machine learns the relationship between the statistics of the specimen test data, the category of the patient's state, and the statistics of the treatment data, and the infectious disease, In the prediction step, input the statistics obtained from the treatment data in addition to the statistics obtained from the specimen test data and the classification result of the patient state data into the prediction model, and predict the possibility that the patient is infected with the infectious disease based on the output result from the prediction model. An infection prediction method characterized by this.

[0152] (Appendix 8) The infection prediction method according to Supplementary Note 7, wherein in the prediction step, as prediction models, a first prediction model that machine-learns the relationship between the statistics of specimen test data, the categories of patient conditions, the statistics of treatment data, and catheter infection, and a second prediction model that machine-learns the relationship between the statistics of specimen test data, the categories of patient conditions, the statistics of treatment data, and urinary tract infection are used; An infection prediction method characterized by the above.

[0153] (Supplementary Note 9) The infection prediction method according to any one of Supplementary Notes 6 to 8, wherein in the data acquisition step, as the specimen test data, the data of the patient's blood test is acquired; in the step of statistical processing of the specimen test data, as the statistics of the specimen test data, any one of the maximum value, the minimum value, and the average value is obtained; An infection prediction method characterized by the above.

[0154] (Supplementary Note 10) The infection prediction method according to any one of Supplementary Notes 6 to 9, wherein in the data acquisition step, as the patient condition data, at least one of the patient's body temperature, blood pressure, pulse, and oxygen saturation is acquired; in the step of classifying the patient condition data, any one of the acquired patient body temperature, blood pressure, pulse, and oxygen saturation is classified into any one of a plurality of preset categories according to the value; An infection prediction method characterized by the above.

[0155] (Supplementary Note 11) A computer is caused to perform a data acquisition step of acquiring specimen test data related to a patient's specimen test and patient condition data indicating the patient's condition, a step of statistical processing of the specimen test data to obtain the statistics of the specimen test data, and a step of classifying the patient condition data into any one of a plurality of preset categories. Input the statistics obtained from the specimen test data and the classification result of the patient status data into a prediction model that is machine learning the relationship between the statistics of the specimen test data, the category of the patient's status, and the infectious disease, and predict the possibility that the patient is infected with the infectious disease based on the output result from the prediction model, a prediction step; A program characterized by causing the above to be executed.

[0156] (Appendix 12) The program according to Appendix 11, wherein In the data acquisition step, further acquire treatment data regarding the treatment performed on the patient, The program further causes the execution of a treatment data statisticalization step of obtaining the statistics of the treatment data, The prediction model machine-learns the relationship between the statistics of the specimen test data, the category of the patient's status, and the statistics of the treatment data, and the infectious disease, In the prediction step, input the statistics obtained from the specimen test data, the classification result of the patient status data, and in addition, the statistics obtained from the treatment data into the prediction model, and predict the possibility that the patient is infected with the infectious disease based on the output result from the prediction model. A program characterized by the above.

[0157] (Appendix 13) The program according to Appendix 12, wherein In the prediction step, as the prediction model, use a first prediction model that machine-learns the relationship between the statistics of the specimen test data, the category of the patient's status, and the statistics of the treatment data, and the catheter infectious disease, and a second prediction model that machine-learns the relationship between the statistics of the specimen test data, the category of the patient's status, and the statistics of the treatment data, and the urinary tract infectious disease. A program characterized by the above.

[0158] (Appendix 14) The program according to any one of Appendices 11 to 13, wherein In the data acquisition step, as the specimen test data, acquire the data of the patient's blood test. In the specimen test data statistical step, as the statistics of the specimen test data, obtain any one of the maximum value, minimum value, and average value. A program characterized by the above.

[0159] (Appendix 15) A program according to any one of Appendices 11 to 14, In the data acquisition step, as the patient state data, acquire at least one of the patient's body temperature, blood pressure, pulse, and oxygen saturation. In the patient state data classification step, classify any one of the acquired patient's body temperature, blood pressure, pulse, and oxygen saturation into any one of a plurality of preset categories according to the value. A program characterized by the above.

Industrial Applicability

[0160] As described above, according to the present invention, it is possible to predict the possibility of a patient's nosocomial infection including the cause of the infection. The present invention is useful in the medical field.

Explanation of Signs

[0161] 100 Infection prediction device 10 Data acquisition unit 20 Pretreatment unit 21 Specimen test data statistical unit 22 Patient state data classification unit 23 Treatment data statistical unit 30 Prediction unit 40 Memory unit 41 Catheter bloodstream infection prediction model 42 Urinary tract infection prediction model 43 Causative bacteria prediction model 44 SSI prediction model 110 Computer 111 CPU 112 Main memory 113 Memory device 114 Input interface 115 Display controller 116 Data reader / writer 117 Communication interface 118 Input device 119 Display device 120 Recording medium 121 Bus 200 Infection prediction device (Embodiment 2) 300 Infection prediction device (Embodiment 3)

Claims

1. A data acquisition unit that acquires specimen test data related to a patient's specimen test and patient status data indicating the status of the patient; A specimen test data statistical unit that calculates the statistics of the specimen test data; A patient status data classification unit that classifies the patient status data into any one of a plurality of preset categories; Input the statistics obtained from the specimen test data and the classification result of the patient status data into a prediction model that is machine learning the relationship between the statistics of the specimen test data, the category of the patient's status, and the infectious disease, predict the possibility that the patient is infected with the infectious disease based on the output result from the prediction model, and output the prediction result; An infection prediction device, characterized by comprising the above.

2. The infection prediction device according to Claim 1, wherein: The data acquisition unit further acquires treatment data related to the treatment performed on the patient; The infection prediction device further comprises a treatment data statistical unit that calculates the statistics of the treatment data; The prediction model is machine learning the relationship between the statistics of the specimen test data, the category of the patient's status, the statistics of the treatment data, and the infectious disease; The prediction unit inputs the statistics obtained from the treatment data in addition to the statistics obtained from the specimen test data and the classification result of the patient status data into the prediction model, and predicts the possibility that the patient is infected with the infectious disease based on the output result from the prediction model. An infection prediction device, characterized by comprising the above.

3. The infection prediction device according to Claim 2, wherein: The prediction unit uses, as the prediction model, a first prediction model that is machine learning the relationship between the statistics of the specimen test data, the category of the patient's status, the statistics of the treatment data, and catheter infectious disease, and a second prediction model that is machine learning the relationship between the statistics of the specimen test data, the category of the patient's status, the statistics of the treatment data, and urinary tract infectious disease. An infection prediction device, characterized by comprising the above.

4. The infection prediction device according to any one of Claims 1 to 3, wherein: The data acquisition unit acquires the data of the patient's blood test as the specimen test data; The specimen test data statistical unit calculates any one of the maximum value, the minimum value, and the average value as the statistics of the specimen test data. An infection prediction device, characterized by comprising the above.

5. The infection prediction device according to any one of Claims 1 to 4, wherein: The data acquisition unit acquires at least one of the patient's body temperature, blood pressure, pulse, and oxygen saturation as the patient state data, The patient state data classification unit classifies any one of the acquired body temperature, blood pressure, pulse, and oxygen saturation of the patient into any one of a plurality of preset categories according to the value, An infection prediction device characterized by this.

6. A method executed by a computer, A data acquisition step of acquiring specimen test data related to a patient's specimen test and patient state data indicating the state of the patient, A specimen test data statisticalization step of obtaining statistics of the specimen test data, A patient state data classification step of classifying the patient state data into any one of a plurality of preset categories, Input the statistics obtained from the specimen test data and the classification result of the patient state data into a prediction model that is machine learning the relationship between the statistics of the specimen test data, the category of the patient's state, and the infectious disease, and predict the possibility that the patient is infected with the infectious disease based on the output result from the prediction model, and output the prediction result, a prediction step, An infection prediction method characterized by having this.

7. The infection prediction method according to claim 6, In the data acquisition step, further acquire treatment data related to the treatment performed on the patient, The infection prediction method further has a treatment data statisticalization step of obtaining statistics of the treatment data, The prediction model has machine learned the relationship between the statistics of the specimen test data, the category of the patient's state, and the statistics of the treatment data, and the infectious disease, In the prediction step, in addition to the statistics obtained from the specimen test data and the classification result of the patient state data, input the statistics obtained from the treatment data into the prediction model, and predict the possibility that the patient is infected with the infectious disease based on the output result from the prediction model. An infection prediction method characterized by this.

8. The infection prediction method according to claim 7, In the prediction step, as the prediction model, a first prediction model that has machine learned the relationship between the statistics of the specimen test data, the category of the patient's state, and the statistics of the treatment data, and catheter infectious disease, and a second prediction model that has machine learned the relationship between the statistics of the specimen test data, the category of the patient's state, and the statistics of the treatment data, and urinary tract infectious disease, are used. An infection prediction method characterized by this.

9. An infection prediction method according to any one of Claims 6 to 8, wherein in the data acquisition step, as the specimen test data, data of a blood test of the patient is acquired; in the specimen test data statisticalization step, as the statistics of the specimen test data, any one of a maximum value, a minimum value, and an average value is obtained; An infection prediction method characterized by the above.

10. An infection prediction method according to any one of Claims 6 to 9, wherein in the data acquisition step, as the patient state data, at least one of the patient's body temperature, blood pressure, pulse, and oxygen saturation is acquired; in the patient state data classification step, any one of the acquired patient body temperature, blood pressure, pulse, and oxygen saturation is classified into any one of a plurality of preset categories according to the value; An infection prediction method characterized by the above.

11. A computer is caused to perform a data acquisition step of acquiring specimen test data related to a patient's specimen test and patient state data indicating the state of the patient; a specimen test data statisticalization step of obtaining statistics of the specimen test data; a patient state data classification step of classifying the patient state data into any one of a plurality of preset categories; inputting the statistics obtained from the specimen test data and the classification result of the patient state data into a prediction model that is machine learning the relationship between the statistics of the specimen test data, the category of the patient's state, and an infectious disease, predicting the possibility that the patient is infected with an infectious disease based on the output result from the prediction model, and outputting a prediction result; A program characterized by causing the above to be executed.

12. A program according to Claim 11, wherein in the data acquisition step, further, treatment data related to a treatment performed on the patient is further acquired; the program further causes a treatment data statisticalization step of obtaining statistics of the treatment data to be executed; the prediction model machine learns the relationship between the statistics of the specimen test data, the category of the patient's state, and the statistics of the treatment data, and an infectious disease; in the prediction step, in addition to the statistics obtained from the specimen test data and the classification result of the patient state data, the statistics obtained from the treatment data are input into the prediction model, and the possibility that the patient is infected with an infectious disease is predicted based on the output result from the prediction model; A program characterized by the above.

13. The program according to claim 12, wherein in the prediction step, as prediction models, a first prediction model that performs machine learning on the relationship between the statistics of specimen test data, the category of the patient's condition, the statistics of treatment data, and catheter infection, and a second prediction model that performs machine learning on the relationship between the statistics of specimen test data, the category of the patient's condition, the statistics of treatment data, and urinary tract infection are used; a program characterized by this.

14. The program according to any one of claims 11 to 13, wherein in the data acquisition step, as the specimen test data, data of the patient's blood test is acquired; in the step of statistically processing the specimen test data, any one of the maximum value, the minimum value, and the average value is obtained as the statistics of the specimen test data; a program characterized by this.

15. The program according to any one of claims 11 to 14, wherein in the data acquisition step, as the patient condition data, at least one of the patient's body temperature, blood pressure, pulse, and oxygen saturation is acquired; in the step of classifying the patient condition data, any one of the acquired patient's body temperature, blood pressure, pulse, and oxygen saturation is classified into any one of a plurality of preset categories according to the value; a program characterized by this.

Images