Anesthetic effect prediction system and program

Abstract

Provided are an anesthetic effect prediction system and a program that provide information that can be easily utilized even by an anesthesiologist with poor knowledge and experience.　An anesthetic effect prediction system 1 comprises: an information acquisition unit 11 that acquires a measurement value of biological information of a target patient to whom an anesthetic has been administered; and an arithmetic processing unit 12 that predicts, on the basis of the measurement value, future variation of the biological information in the target patient.

Description

Anesthetic effect prediction system and program

[0001] The present invention relates to an anesthesia effect prediction system and program for predicting the effect of anesthesia.

[0002] Anesthesiologists not only administer anesthetic drugs to patients to induce a coma before surgery, but also assess the patient's condition from measurements taken during surgery, such as the patient's heart rate, blood pressure, and blood flow, and then administer additional anesthetic drugs, painkillers, and hypertensives to stabilize the patient's condition.

[0003] For example, Patent Document 1 proposes a system that calculates an index value of a patient's nociceptive stimulus response level from the patient's measurements. By monitoring the index value calculated by such a system in addition to the above measurements, anesthesiologists can more accurately grasp the patient's condition.

[0004] JP 2018-130542 A

[0005] However, in order for an anesthesiologist to refer to the patient's measurements and index values ​​and provide appropriate treatment to the patient, they need the knowledge and experience to properly utilize this information, so there are only a limited number of anesthesiologists who can fully utilize a system such as that proposed in Patent Document 1.

[0006] The present invention has been made to solve the above-mentioned problems, and provides an anesthetic effect prediction system and program that provide information that can be easily used even by anesthesiologists with little knowledge or experience.

[0007] In order to achieve the above-mentioned objectives, the anesthetic effect prediction system disclosed below comprises an information acquisition unit that acquires measurements of the biometric information of a target patient who has been administered an anesthetic, and a calculation processing unit that predicts future changes in the biometric information of the target patient based on the measurements.

[0008] The above-described system for predicting the effect of anesthesia can obtain future biological information about a target patient. Therefore, even an anesthesiologist with so little knowledge and experience that he or she is unable to predict future biological information from the current biological information of a target patient can obtain the future biological information about the target patient and easily determine the treatment that should be performed now. Therefore, this system for predicting the effect of anesthesia can provide information that can be easily utilized even by an anesthesiologist with little knowledge and experience.

[0009] Fig. 1 is a block diagram showing an example of the configuration of an anesthetic effect prediction system 1 according to an embodiment of the present invention. Fig. 2 is a graph showing an example of a method for outputting a prediction result by the anesthetic effect prediction system 1 according to an embodiment of the present invention. Fig. 3 is a graph showing another example of a method for outputting a prediction result by the anesthetic effect prediction system 1 according to an embodiment of the present invention.

[0010] An embodiment of the present invention will be described below with reference to the drawings. The present invention is not limited to the following embodiment, and appropriate design modifications can be made within the scope of the configuration of the present invention. In the following description, the same reference numerals are used in different drawings for identical parts or parts having similar functions, and repeated description thereof will be omitted. The configurations described in the embodiment and modified examples may be combined or modified as appropriate. To facilitate understanding of the description, the drawings referred to below show simplified or schematic configurations, or some of the configurations may be omitted.

[0011] First, an example of the configuration of an anesthetic effect prediction system 1 according to an embodiment of the present invention will be described with reference to the drawings. Fig. 1 is a block diagram showing an example of the configuration of an anesthetic effect prediction system 1 according to an embodiment of the present invention. Note that, hereinafter, a patient who undergoes surgery in a coma state after being administered an anesthetic and whose anesthesia effect is predicted by the anesthetic effect prediction system 1 will be referred to as a "target patient."

[0012] As shown in FIG. 1, the anesthetic effect prediction system 1 includes an information acquisition unit 11, a calculation processing unit 12, an output unit 13, and a storage unit 14.

[0013] The information acquisition unit 11 is composed of an interface into which measurement values ​​are input from various measuring devices (sphygmomanometer, electrocardiograph, thermometer, pulse oximeter, etc.) attached to the target patient, and communication devices that receive these measurement values ​​via wireless communication.

[0014] The arithmetic processing unit 12 is configured by an arithmetic processing device such as a CPU (Central Processing Unit).

[0015] The output unit 13 is composed of a device that outputs visually, such as a display device that displays images or a segment display that displays alphanumeric characters, or a device that outputs auditorily, such as a speaker that emits voice or warning sounds.

[0016] The storage unit 14 is configured by, for example, a non-volatile storage device such as a hard disk drive (HDD) or a solid state drive (SSD), or a volatile storage device such as a random access memory (RAM).

[0017] The information acquisition unit 11, the arithmetic processing unit 12, and the storage unit 14 may be configured as part of a personal computer or a server computer. The anesthesia effect prediction system 1 may also be configured as part of an anesthesia management system that records and manages biological information of a patient undergoing surgery. The arithmetic processing unit 12 and the storage unit 14 may also be configured as part of a server or the like installed in a location remote from the information acquisition unit 11 and the output unit 13, and may exchange information with the information acquisition unit 11 and the output unit 13 via a communication device.

[0018] The storage unit 14 stores a trained model 141 and simulation information 142. In addition to this information, the storage unit 14 also stores measurement values ​​acquired by the information acquisition unit 11 and programs necessary for the operation of the anesthesia effect prediction system 1 (and further the anesthesia management system).

[0019] The trained model 141 is obtained by machine learning a model such as a neural network model using time-series data of the biological information of patients who have been administered anesthetics as training data. The training data can be data on multiple patients recorded by an anesthesia management system or the like in past surgeries (e.g., the administration history of drugs such as anesthetics, vasopressors, and analgesics, and time-series data of the patient's biological information). Any model, such as a deep neural network (DNN), a convolutional neural network (CNN), or a recurrent neural network (RNN), can be used as the model.

[0020] The simulation information 142 is data for performing a simulation using a pharmacokinetic model such as a multi-compartment model, which is introduced in "Simulation Using a Pharmacokinetic Multi-Compartment Model and Clinical Anesthesia" (authors: Kenichi Masui and Tomie Kazama), Vol. 26, No. 3, pp. 145-151 of "Simulation" published by the Japan Society for Simulation Technology (to calculate the concentration of an anesthetic agent in the body of a subject patient and its changes).

[0021] Next, the operation of the anesthesia effect prediction system 1 according to an embodiment of the present invention will be described. First, in the anesthesia effect prediction system 1, measurement values ​​from a measuring device attached to a target patient are acquired via the information acquisition unit 11. Note that if the anesthesia effect prediction system 1 is configured as part of an anesthesia management system that records and manages the biological information of a target patient during surgery, the measurement values ​​input to the anesthesia management system 1 will also be used in the anesthesia effect prediction system 1. Furthermore, the information acquisition unit 11 periodically acquires measurement values, for example, at a predetermined cycle when the measuring device operates.

[0022] The calculation processing unit 12 inputs the measurement values ​​input via the information acquisition unit 11 to the trained model 141, and predicts future changes in the biological information of the target patient based on the obtained output results. At this time, the calculation processing unit 12 may input, for example, measurement values ​​within a certain period of time in the past (e.g., the past few minutes) to the trained model 141.

[0023] In addition, the calculation processing unit 12 calculates the simulation results of the anesthetic concentration in the target patient based on the measurement value simulation information 142 input via the information acquisition unit 11, and predicts future fluctuations in the biological information based on this simulation result.

[0024] For example, the calculation processing unit 12 may calculate a prediction result of future changes in the target patient's biological information by using the results of both the output of the trained model 141 and the simulation results in parallel, such as by combining or comparing them.

[0025] Furthermore, for example, the calculation processing unit 12 may input simulation results together with measured values ​​into the trained model 141 to obtain output results, and use the respective results serially. In this case, simulation results may be calculated for each of multiple patient data recorded by an anesthesia management system or the like in past surgeries, and these may be used as training data to perform machine learning on the model. The effect of anesthesia is more strongly correlated with the concentration of the anesthetic in the body than with the amount or timing of administration. Therefore, by focusing on the simulation results of the concentration and using them as training data to perform machine learning to obtain the trained model 141, and inputting the simulation results of the anesthetic concentration in the target patient into this trained model 141, future changes in the target patient's biological information can be more accurately determined.

[0026] The output unit 13 outputs the output results obtained by the calculations of the calculation processing unit 12 in a manner that can be recognized by humans, such as a visual method such as an image or an auditory method such as a sound. An example of this output method will be described with reference to Fig. 2. Fig. 2 is a graph showing an example of a method for outputting prediction results by the anesthetic effect prediction system 1 according to an embodiment of the present invention.

[0027] 2 illustrates an example in which the arithmetic processing unit 12 predicts future changes in the biological information of a target patient, outputs the prediction results as time-series information, and the output unit 13 displays the time-series information as an image. In the image shown in FIG. 2, for the time period before the present, the measurement values ​​of the target patient are displayed as a time-series graph 131, and for the time period after the present, the prediction results by the arithmetic processing unit 12 are displayed as a time-series graph 132. In this way, when the prediction results are output as time-series information, even an anesthesiologist with little knowledge or experience can easily determine not only the need for treatment for the target patient but also the urgency (e.g., how much time remains until the patient becomes critically ill).

[0028] As described above, the anesthetic effect prediction system 1 according to an embodiment of the present invention can obtain future biological information of a target patient. Therefore, even an anesthesiologist with so little knowledge and experience that he or she is unable to predict future biological information from the current biological information of a target patient can obtain the target patient's future biological information and easily determine the treatment that should be performed now. Therefore, the anesthetic effect prediction system 1 can provide information that can be easily utilized even by an anesthesiologist with little knowledge and experience.

[0029] <Modifications, etc.> The above-described embodiments are merely examples for carrying out the present invention. The present invention is not limited to the above-described embodiments, and can be carried out by appropriately modifying the above-described embodiments within the scope of the spirit of the present invention.

[0030] For example, in the above-described embodiment, the memory unit 14 has been described as storing both the trained model 141 and the simulation information 142, but it may store only one of them, or it may store data related to a prediction method other than these.

[0031] Furthermore, for example, the trained model 141 may be a model that has been machine-learned based on not only the patient's measurements but also attribute information (e.g., gender, age, height, weight, etc.). In this case, the calculation processing unit 12 may input the attribute information of the target patient into the trained model 141 to obtain a prediction result. The effects of anesthesia vary depending on gender, age, and physique, so by distinguishing between these factors and having the trained model input, it is possible to accurately predict the target patient's biological information.

[0032] In the above-described embodiment, the measured and predicted biological information may include blood pressure. In this case, the anesthetic effect prediction system 1 can predict blood pressure, which is an important index in the management of anesthesia for a target patient.

[0033] 2, the calculation processing unit 12 may output information indicating that treatment is required for the target patient in addition to (or instead of) the time-series data. In this case, even an anesthesiologist with so little knowledge and experience that it is difficult for him or her to determine the necessary treatment can recognize that treatment is required.

[0034] 2 illustrates an output image that combines a graph 131 of the measurement values ​​of the target patient and a graph 132 of the prediction result, the output unit 13 may output an output image other than this. For example, as shown in Fig. 3, a graph 131 of the measurement values ​​of the target patient's biological information may be displayed superimposed with a past prediction result 133 of the biological information. For example, the prediction result 133 at a certain point in time in the past is the biological information of the target patient at the certain point in time that was predicted a predetermined time before (e.g., several minutes before) from the certain point in time.

[0035] The calculation processing unit 12 outputs time-series information in which past prediction results for the target patient's biological information are superimposed on the measured values ​​of the biological information, so as to obtain the output result shown in Fig. 3. This makes it possible to clearly indicate whether the measured values ​​for the target patient and the prediction results generally match (whether the prediction accuracy is sufficiently high). Therefore, the anesthesiologist who is the user can easily determine whether the output results of the anesthetic effect prediction system 1 are reliable.

[0036] The above-described anesthetic effect prediction system can be explained as follows.

[0037] The anesthetic effect prediction system includes an information acquisition unit that acquires measurements of biological information of a target patient who has been administered an anesthetic, and a calculation processing unit that predicts future changes in the biological information of the target patient based on the measurements (first configuration). This configuration allows the target patient's future biological information to be obtained. Therefore, even an anesthesiologist with limited knowledge and experience who is unable to predict future biological information from the target patient's current biological information can obtain the target patient's future biological information, thereby easily determining the treatment that should be performed. Therefore, this anesthetic effect prediction system can provide information that even an anesthesiologist with limited knowledge and experience can easily use.

[0038] In the first configuration, the system may further include a storage unit that stores a trained model that has been machine-learned using time-series data of the biological information of a patient who has been administered an anesthetic as training data, and the calculation processing unit may predict future fluctuations in the biological information of the target patient based on an output result obtained by inputting the measurement values ​​of the target patient into the trained model (second configuration). With this configuration, the biological information of the target patient can be predicted with high accuracy.

[0039] In addition, in the second configuration, the trained model may be machine-learned using time-series data of the patient's biological information and attribute information of the patient as training data, the information acquisition unit acquires the attribute information of the target patient, and the calculation processing unit may predict future fluctuations in the biological information of the target patient based on an output result obtained by inputting the measurement values ​​and the attribute information of the target patient into the trained model (third configuration). With this configuration, predictions can be made taking the attributes of the target patient into consideration, thereby making it possible to predict the biological information of the target patient with even greater accuracy.

[0040] In any one of the first to third configurations, the arithmetic processing unit may predict future fluctuations in the blood pressure of the subject patient (fourth configuration). With this configuration, it is possible to predict blood pressure, which is an important index in the anesthesia management of the subject patient.

[0041] In any one of the first to fourth configurations, the arithmetic processing unit may predict future changes in the biological information of the target patient and output the prediction results as time-series information (fifth configuration). With this configuration, even an anesthesiologist with little knowledge or experience can easily determine the necessity and urgency of treatment for the target patient.

[0042] In any one of the first to fifth configurations, the arithmetic processing unit may output time-series information in which past prediction results for the biological information of the target patient are superimposed on the measured values ​​of the biological information (sixth configuration). This configuration makes it possible to clearly indicate whether the measured values ​​of the target patient and the prediction results generally match (whether the prediction accuracy is sufficiently high). Therefore, the anesthesiologist, who is the user, can easily determine whether the output results of the anesthetic effect prediction system are reliable.

[0043] In any one of the first to sixth configurations, the arithmetic processing unit may output information for notifying a user when the predicted biological information of the target patient indicates that treatment is required for the target patient at present or in the future (seventh configuration). With this configuration, even an anesthesiologist with so little knowledge and experience that it is difficult to determine the necessary treatment can recognize that treatment is required.

[0044] In any one of the first to seventh configurations, the information acquisition unit may acquire information about the anesthetic administered to the target patient, and the arithmetic processing unit may calculate a simulation result of the concentration of the anesthetic in the target patient based on the information about the anesthetic administered to the target patient, and predict future fluctuations in the biological information of the target patient based on the simulation result (eighth configuration). With this configuration, it is possible to accurately predict the biological information of the target patient.

[0045] The first configuration may further include a memory unit that stores a trained model that has been machine-learned using time-series data of the body information of a patient who has been administered an anesthetic and a simulation result of the concentration of the anesthetic in the patient as training data, wherein the information acquisition unit acquires information about the anesthetic administered to the target patient, the arithmetic processing unit calculates the simulation result of the anesthetic concentration in the target patient based on the information about the anesthetic administered to the target patient, and the arithmetic processing unit predicts future changes in the biological information of the target patient based on an output result obtained by inputting the measured values ​​of the target patient and the simulation result of the anesthetic concentration in the target patient into the trained model (ninth configuration). With this configuration, focusing on the simulation result of the anesthetic concentration in the body, which has a strong correlation with the effect of anesthesia, the trained model is trained using the simulation result as training data to obtain a trained model, and the simulation result of the anesthetic concentration in the target patient is input into the trained model, thereby more accurately determining future changes in the biological information of the target patient.

[0046] In addition, the program used in the above-mentioned anesthetic effect prediction system causes a computer to perform the following processes: acquiring measurements of the biometric information of a target patient who has been administered an anesthetic; and predicting future changes in the biometric information of the target patient based on the measurements (tenth configuration).

[0047] 1...anesthetic effect prediction system, 11...information acquisition unit, 12...arithmetic processing unit, 13...output unit, 14...storage unit, 141...trained model, 142...simulation information

Claims

1. An anesthetic effect prediction system comprising: an information acquisition unit that acquires measurements of the biological information of a target patient who has been administered an anesthetic; and a calculation processing unit that predicts future fluctuations in the biological information of the target patient based on the measurements.

2. An anesthetic effect prediction system as described in claim 1, further comprising a memory unit that stores a trained model that has been machine-learned using time series data of the biometric information of a patient who has been administered an anesthetic as training data, and the calculation processing unit predicts future fluctuations in the biometric information of the target patient based on the output results obtained by inputting the measurement values ​​of the target patient into the trained model.

3. The anesthetic effect prediction system described in claim 2, wherein the trained model is machine-learned using time series data of the patient's biometric information and the patient's attribute information as training data, the information acquisition unit acquires the attribute information of the target patient, and the calculation processing unit predicts future fluctuations in the biometric information of the target patient based on the output results obtained by inputting the measurement values ​​and the attribute information of the target patient into the trained model.

4. The system for predicting an effect of an anesthetic according to claim 1, wherein the calculation processing unit predicts future fluctuations in the blood pressure of the subject patient.

5. The anesthetic effect prediction system according to claim 1, wherein the calculation processing unit predicts future fluctuations in the biological information of the subject patient and outputs the prediction results as time-series information.

6. The anesthetic effect prediction system of claim 1, wherein the calculation processing unit outputs time-series information in which the measurement values ​​of the biological information of the target patient are superimposed with past prediction results of the biological information.

7. The anesthetic effect prediction system of claim 1, wherein the calculation processing unit outputs information to notify the user if the predicted biological information of the target patient indicates that treatment is required for the target patient now or in the future.

8. An anesthetic effect prediction system as described in any one of claims 1 to 7, wherein the information acquisition unit acquires information about the anesthetic administered to the target patient, and the calculation processing unit calculates a simulation result of the concentration of the anesthetic in the target patient based on the information about the anesthetic administered to the target patient, and predicts future fluctuations in the biological information of the target patient based on the simulation result.

9. An anesthetic effect prediction system as described in claim 1, further comprising a memory unit that stores a trained model that has been machine-learned using time series data of the body information of a patient who has been administered an anesthetic and simulation results of the concentration of the anesthetic in the patient as training data, wherein the information acquisition unit acquires information about the anesthetic administered to the target patient, the calculation processing unit calculates simulation results of the concentration of the anesthetic in the target patient based on the information about the anesthetic administered to the target patient, and the calculation processing unit predicts future fluctuations in the biological information of the target patient based on output results obtained by inputting the measurement values ​​of the target patient and the simulation results of the concentration of the anesthetic in the target patient into the trained model.

10. A program that causes a computer to perform the following processes: acquiring measurements of vital signs of a subject patient who has been administered an anesthetic; and predicting future fluctuations in the vital signs of the subject patient based on the measurements.

Images