Program, information processing method, and information processing device

Abstract

In the present invention, a program causes a computer to execute processing for: acquiring time series data indicating the glucose level and the lactate level of a patient in a time series; estimating the blood glucose level of the patient after a prescribed period of time on the basis of the time series data of the glucose level and the lactate level; and determining a dosage of insulin to the patient on the basis of the estimated blood glucose level. Preferably, the program estimates the blood glucose level after a prescribed period of time by inputting the acquired time-series data of the glucose level and the lactate value into a model trained so as to estimate the blood glucose level after the prescribed period of time when the time-series data of the glucose level and the lactate level is input.

Description

Program, information processing method, and information processing apparatus.

[0001] This invention relates to a program, an information processing method, and an information processing apparatus.

[0002] There are technologies that control insulin dosage based on a patient's glucose levels. For example, Patent Document 1 discloses a closed-loop blood glucose control system that uses a physiological model to determine the amount of insulin to inject from glucose measurements and provides an insulin supply control signal to an insulin supply device.

[0003] Patent No. 7290624

[0004] One aspect of this is the aim to provide a program or the like that can appropriately determine the amount of insulin administered to a patient.

[0005] (1) The program obtains time-series data showing the patient's glucose and lactate levels over time, estimates the patient's blood glucose level after a predetermined time based on the time-series data of glucose and lactate levels, and causes the computer to perform the process of determining the amount of insulin to be administered to the patient based on the estimated blood glucose level.

[0006] (2) The program described in (1) above estimates the blood glucose level after a predetermined time by inputting the acquired time series data of glucose and lactate into a model that has been trained to estimate the blood glucose level after a predetermined time when time series data of glucose and lactate levels are input.

[0007] (3) The program described in (2) above comprises a first network that estimates the difference from the current value of the lactate value after a predetermined time when time-series data of the glucose value and lactate value are input, and a second network that estimates the difference from the current value of the blood glucose value after a predetermined time when time-series data of the glucose value and lactate value and the difference from the current value of the lactate value after a predetermined time are input. The program estimates the difference from the current value of the lactate value after a predetermined time by inputting the acquired time-series data of the glucose value and lactate value into the first network, and estimates the difference from the current value of the blood glucose value after a predetermined time by inputting the acquired time-series data of the glucose value and lactate value and the estimated difference from the current value of the lactate value after a predetermined time into the second network.

[0008] (4) The program described in any of (1) to (3) above further acquires time-series data showing the actual and planned amounts of insulin administered to the patient in a time series, and estimates the blood glucose level after a predetermined time based on the glucose value, lactate value, and administration time-series data.

[0009] (5) The program described in any of (1) to (4) above further acquires time-series data showing the actual and planned values ​​of the patient's eating speed in a time series, and estimates the blood glucose level after a predetermined time based on the glucose value, lactate value, and time-series data of eating speed.

[0010] (6) The program described in any of (1) to (5) above increases the insulin dose from the current value to a predetermined value if the estimated blood glucose level exceeds the first threshold, and decreases the insulin dose from the current value to a predetermined value if the estimated blood glucose level is below the second threshold.

[0011] (7) The information processing method involves a computer that acquires time-series data showing the patient's glucose and lactate values ​​in a time series, estimates the patient's blood glucose level after a predetermined time based on the time-series data of glucose and lactate values, and determines the amount of insulin to be administered to the patient based on the estimated blood glucose level.

[0012] (8) The information processing device is an information processing device comprising a control unit, wherein the control unit acquires time-series data showing the patient's glucose and lactate values ​​in time series, estimates the patient's blood glucose level after a predetermined time based on the time-series data of glucose and lactate values, and determines the amount of insulin to be administered to the patient based on the estimated blood glucose level.

[0013] In one respect, it allows for the appropriate determination of the insulin dosage to be administered to the patient.

[0014] This is an explanatory diagram showing an example configuration of an insulin administration system. This is a block diagram showing an example server configuration. This is a block diagram showing an example terminal configuration. This is a diagram showing an example estimation model configuration. This is a diagram showing an example second estimation model configuration. This is a flowchart showing the procedure for generating the estimation model. This is a flowchart showing the procedure for insulin administration control processing.

[0015] The present invention will be described in detail below with reference to the drawings illustrating its embodiments. (Embodiment) Figure 1 is an explanatory diagram showing an example of the configuration of an insulin administration system. In this embodiment, an insulin administration system is described that estimates the blood glucose level of a diabetic patient (for example, a patient with type 1 diabetes) based on the glucose level and lactate level of the patient, and controls (determines) the amount of insulin administered to the patient. Since the blood glucose level is linked to the glucose concentration in the interstitial fluid, the blood glucose level may be equivalent to the glucose level in the interstitial fluid. The insulin administration system includes a server 1, a terminal 2, a sensor 3, and an infusion device 4. The server 1 and the terminal 2 are connected via a network N such as the Internet.

[0016] Sensor 3 is a sensor that measures glucose levels (e.g., glucose concentration) in the subcutaneous tissue (interstitial fluid) of the patient, and is a sensor known as a CGM (Continuous Glucose Monitoring) sensor. Sensor 3 periodically measures glucose levels (e.g., once every 1 minute to a few minutes) and transmits the measured values ​​to terminal 2.

[0017] In this embodiment, the sensor 3 measures not only glucose levels but also lactate levels (e.g., lactate concentration). As will be described later, by measuring not only glucose levels but also lactate levels, the patient's exercise status can be understood and the insulin dosage can be appropriately controlled. For example, the sensor 3 is provided with a sensor region in which lactose dehydrogenase is immobilized on a microelectrode, so that the subcutaneous lactate concentration is measured periodically (e.g., once every 1 minute to a few minutes), and the measured value is transmitted to the terminal 2.

[0018] The infusion device 4 is a so-called insulin pump that continuously infuses insulin into the patient's body. The terminal 2 controls the amount (infusion rate) of insulin by outputting a control signal to the infusion device 4 that instructs the amount of insulin to be administered.

[0019] Terminal 2 is an information processing terminal capable of various information processing and information transmission / reception, such as a smartphone or tablet. Terminal 2 periodically acquires glucose and lactate values ​​from sensor 3 and estimates the patient's blood glucose level after a predetermined time (for example, 1 hour) from the time-series data. Terminal 2 determines the insulin dosage according to the estimated blood glucose level and outputs a control signal to the infusion device 4.

[0020] Server 1 is an information processing device capable of various information processing functions, and is a computer that generates (constructs) an estimation model 50 (see Figure 4) used by terminal 2 to estimate blood glucose levels using machine learning. By learning predetermined training data, Server 1 generates an estimation model 50 that estimates (outputs) blood glucose levels after a predetermined time when time-series data such as glucose values ​​and lactate values ​​are input. The generated estimation model 50 data is installed on terminal 2.

[0021] Figure 2 is a block diagram showing an example configuration of Server 1. Server 1 comprises a control unit 11, a main memory unit 12, a communication unit 13, and an auxiliary storage unit 14. The control unit 11 is a processor such as one or more CPUs (Central Processing Units), MPUs (Micro-Processing Units), or GPUs (Graphics Processing Units), and performs various information processing by reading and executing the program P1 stored in the auxiliary storage unit 14. The main memory unit 12 is a temporary storage area such as SRAM (Static Random Access Memory) or DRAM (Dynamic Random Access Memory), and temporarily stores data necessary for the control unit 11 to perform calculations. The communication unit 13 is a communication module for performing communication-related processing, and transmits and receives information with the outside. The auxiliary storage unit 14 is a non-volatile storage area such as a hard disk or large-capacity memory, and stores the program P1 (program product) and other data necessary for the control unit 11 to perform processing.

[0022] The auxiliary storage unit 14 may be an external storage device connected to the server 1. Furthermore, the server 1 may be a multi-computer system consisting of multiple computers, or it may be a virtual machine virtually constructed by software.

[0023] Furthermore, in this embodiment, the server 1 is not limited to the above configuration and may include, for example, an input unit for receiving operation input, a display unit for displaying images, etc. Also, the server 1 may be equipped with a reading unit for reading portable storage media 1a such as CD (Compact Disk)-ROM or DVD (Digital Versatile Disc)-ROM, and may read and execute the program P1 from the portable storage media 1a.

[0024] Figure 3 is a block diagram showing an example configuration of terminal 2. Terminal 2 comprises a control unit 21, a main memory unit 22, a communication unit 23, a display unit 24, an input unit 25, and an auxiliary storage unit 26. The control unit 21 is one or more processors such as CPUs, and performs various information processing by reading and executing the program P2 stored in the auxiliary storage unit 26. The main memory unit 22 is a temporary storage area such as RAM, and temporarily stores data necessary for the control unit 21 to perform calculation processing. The communication unit 23 is a communication module for performing communication-related processing, and transmits and receives information with the outside. The display unit 24 is a display screen such as a liquid crystal display, and displays images. The input unit 25 is an operation interface such as a touch panel, and accepts operation input.

[0025] The auxiliary storage unit 26 is a non-volatile storage area such as a hard disk, and stores the program P2 (program product) and other data necessary for the control unit 21 to execute processing. The auxiliary storage unit 26 also stores the estimation model 50. The estimation model 50 is a machine learning model that has been trained on predetermined training data, and is a model that estimates the blood glucose level of a patient after a predetermined time when time-series data such as glucose value and lactate value is input.

[0026] The terminal 2 may also be equipped with a reading unit to read the portable storage medium 2a, and may read and execute the program P2 from the portable storage medium 2a.

[0027] Figure 4 shows an example of the configuration of the estimation model 50. Figure 4 illustrates the network structure of the estimation model 50 for estimating a patient's blood glucose level after a predetermined time (for example, after 1 hour). The outline of this embodiment will be described below.

[0028] As is widely known, when diabetic patients exercise, they may experience hypoglycemia due to increased insulin sensitivity or the consumption of glucose in their bodies. In such cases, it may be necessary to adjust the insulin dosage.

[0029] Blood lactate levels (lactate concentration) increase in response to exercise. Therefore, in this embodiment, by continuously monitoring not only glucose levels (glucose concentration) but also lactate levels, the patient's exercise status can be understood without the patient having to input the data themselves, and the appropriate insulin dosage can be determined.

[0030] Specifically, terminal 2 uses an estimation model 50, which has been trained to estimate blood glucose levels when glucose levels, lactate levels, etc., are input, to estimate the patient's blood glucose level after a predetermined time. Terminal 2 compares the estimated blood glucose level with a threshold and adjusts (determines) the insulin dosage.

[0031] In this embodiment, a machine learning model (estimation model 50) is used as the algorithm for estimating blood glucose levels, but this embodiment is not limited to this. Terminal 2 may estimate blood glucose levels using algorithms other than machine learning models (for example, models using artificial intelligence or the Hovorka model described later).

[0032] The estimation model 50 is a machine learning model that has been trained on predetermined training data, and is a model that estimates the patient's blood glucose level after a predetermined time period, taking at least time-series data showing the patient's glucose and lactate levels in time series as input. For example, a neural network related to LSTM (Long-Short Term Memory) can be used as the estimation model 50.

[0033] Note that the estimated model 50 may be a neural network other than LSTM, and may be a machine learning model other than a neural network, such as a decision tree or SVM (Support Vector Machine).

[0034] The estimation model 50 according to this embodiment takes glucose and lactate values, as well as the amount of insulin administered to the patient (basal administration rate and bolus administration rate) and the patient's eating rate as inputs. Specifically, as shown in the input layer labeled "input1" in Figure 4, the actual values ​​of glucose, lactate, insulin administration, and eating rate from a predetermined time ago (e.g., 6 hours ago) to the present are inputs. Also, as shown in the input layer labeled "input2" in Figure 4, the planned values ​​of insulin administration and eating rate from the present to a predetermined time later (e.g., 1 hour later) are inputs.

[0035] As shown on the left and right sides of Figure 4, the estimation model 50 includes a first network 51 (right side of Figure 4) that estimates the difference in lactate levels from the current value after a predetermined time, and a second network 52 (left side of Figure 4) that estimates the difference in blood glucose levels from the current value after a predetermined time. Parameters entered into "input1" and "input2" are input to the first network 51 and the second network 52, respectively, via LSTM blocks. The difference in lactate levels output from the first network 51 is also input to the second network 52.

[0036] Server 1 generates an estimation model 50 using predetermined training data. The training data consists of time-series data of continuous glucose values, continuous lactate values, insulin dosage (basal administration rate and bolus administration rate), and eating rate, with the correct values ​​of the difference between the current lactate value and the current blood glucose value after a predetermined time period labeled.

[0037] Server 1 first generates a first network 51 using training data. Server 1 inputs time-series data of training glucose values, lactate values, insulin dosage, and eating speed into the first network 51 to estimate the difference in lactate value from the current value after a predetermined time. Server 1 compares the estimated difference value with the ground truth value and optimizes parameters such as the weights between neurons in the first network 51 so that the two approximate each other.

[0038] Next, the server 1 generates the second network 52 using the training data. The server 1 inputs the time-series data of the glucose value, lactic acid value, insulin dosage, and eating speed for training into the trained first network 51 to estimate the difference from the current value of the lactic acid value after a predetermined time. Then, the server 1 inputs the time-series data of the glucose value, lactic acid value, insulin dosage, and eating speed for training and the difference value of the lactic acid value output from the first network 51 into the second network 52 to estimate the difference from the current value of the blood glucose value after a predetermined time. The server 1 compares the estimated difference value with the correct value and optimizes the parameters such as the weights between neurons in the second network 52 so that the two are approximated.

[0039] As described above, the server 1 generates an estimation model 50 composed of a first network 51 for estimating the lactic acid value after a predetermined time and a second network 52 for estimating the blood glucose value after a predetermined time. The data of the generated estimation model 50 is installed in the terminal 2 that controls the dosage of insulin.

[0040] The terminal 2 periodically acquires and records the measured values of the glucose value and lactic acid value of the patient from the sensor 3. The terminal 2 estimates the blood glucose value of the patient after a predetermined time from the measured values of the glucose value and lactic acid value every predetermined period (for example, 5 minutes) and controls the dosage of insulin.

[0041] Specifically, the terminal 2 reads out the glucose value and lactic acid value for the most recent predetermined time (for example, 6 hours). In addition, the terminal 2 reads out the actual values of the insulin dosage (basal administration rate, bolus administration rate) and eating speed for the most recent predetermined time and the planned values of the insulin dosage and eating speed from now until after a predetermined time (for example, 1 hour later). The planned values of the insulin dosage and eating speed are predetermined.

[0042] The terminal 2 estimates the blood glucose level of the patient after a predetermined time by inputting the time-series data of the glucose level, lactic acid level, insulin dosage, and eating speed into the estimation model 50. Specifically, the terminal 2 inputs the actual values of the glucose level, lactic acid level, insulin dosage, and eating speed from a predetermined time ago to the present into "input1", and inputs the planned values of the insulin dosage and eating speed from the present to after the predetermined time into "input2".

[0043] By inputting these time-series data, the terminal 2 estimates, from the first network 51, the difference from the current value of the lactic acid level after a predetermined time. The terminal 2 inputs the time-series data of the glucose level, lactic acid level, insulin dosage, and eating speed, and the difference value of the lactic acid level output from the first network 51 into the second network 52 to estimate the difference from the current value of the blood glucose level after a predetermined time. The terminal 2 estimates the blood glucose level after a predetermined time based on the estimated difference value and the current value of the blood glucose level.

[0044] The terminal 2 determines the dosage of insulin based on the estimated blood glucose level. When the blood glucose level exceeds the first threshold value (for example, 180 mg / dL), the terminal 2 increases the insulin dosage (for example, the basal infusion rate) by a predetermined value (for example, 0.1 U / h). Also, when the blood glucose level is below the second threshold value (for example, 100 mg / dL), the terminal 2 decreases the insulin dosage by a predetermined value (for example, 0.1 U / h).

[0045] The terminal 2 generates a control signal for instructing the determined insulin dosage and outputs it to the infusion device 4. The infusion device 4 varies (increases or decreases) the insulin dosage according to the control signal.

[0046] Thus, in this embodiment, not only the glucose level but also the lactic acid level is considered to estimate the blood glucose level and determine the insulin dosage. Thereby, the exercise state of the patient can be grasped and the insulin dosage can be suitably controlled. Particularly in this embodiment, the lactic acid level after a predetermined time is estimated by the first network 51 of the estimation model 50 and used as an input to the second network 52, whereby the blood glucose level can be suitably estimated.

[0047] The following section describes a comparative example of estimating blood glucose levels based on glucose and lactate levels, and estimating blood glucose levels without using lactate levels.

[0048] Figure 5 shows an example of the configuration of the second estimation model 60. Figure 5 illustrates the network structure of the second estimation model 60, which was constructed for comparison with the estimation model 50 described above.

[0049] The second estimation model 60, like estimation model 50, is a neural network related to LSTM. The main difference between estimation model 50 and the second estimation model 60 is that the second estimation model 60 does not use lactate value (lactate concentration) as input, but estimates the blood glucose level (difference from the current value) after a predetermined time from glucose value (glucose concentration), insulin dosage (basal administration rate, bolus administration rate), and meal rate. Also, because lactate value is not used, the second estimation model 60 does not estimate the lactate value after a predetermined time.

[0050] The second estimation model 60 accepts input in "input1" for continuous glucose values, insulin dosage, and eating speed from a predetermined time before (e.g., 6 hours before) to the present, and in "input2" for planned insulin dosage and eating speed from the present to a predetermined time after (e.g., 1 hour after). Based on this time-series data, the second estimation model 60 estimates the difference in blood glucose levels after the predetermined time from the current value. Based on the difference in blood glucose levels output by the second estimation model 60, the blood glucose level after the predetermined time can be estimated.

[0051] The inventor of the present application conducted an experiment by simulation to compare the performance of the estimation model 50 and the second estimation model 60. Specifically, as a mathematical model simulating glucose metabolism in the human body, an experiment was conducted using a modified model (hereinafter referred to as the modified Hovorka model) of the mathematical model described in the following non-patent literature. [Non-patent literature] Charrise M. Ramkissoon et al., “A Model of Glucose-Insulin-Pramlintide Pharmacokinetics and Pharmacodynamics in Type I Diabetes”, [online], [searched on October 3, 2024], Internet <https: / / www.ncbi.nlm.nih.gov / pmc / articles / PMC4455443 / >

[0052] The Hovorka model is a model that simulates the influence of insulin administration on blood glucose levels in the body. The Hovorka model (modified Hovorka model) according to the present embodiment is defined by the following mathematical formulas (1) to (3).

[0053]

[0054] Q １ represents the glucose concentration in the blood, Q ２ represents the glucose concentration in the muscle, x １ represents the rate constant of glucose transfer from the blood to the muscle, x ２ represents the glucose consumption rate constant of the muscle, k １２ represents the rate constant of glucose transfer from the muscle to the blood, F ０１ represents the glucose consumption independent of insulin, F Ｒ represents the glucose excretion by urine, EGP represents the glucose production in the body, U Ｇ represents the glucose intake from the external (diet), Ψ represents the exercise state (glycolysis enhancement ratio), L represents the lactic acid concentration in the blood, K ｅｌΨ represents the lactate metabolism rate constant, and α represents a constant. Sigmoid(Ψ) is a monotonically increasing sigmoid function with the exercise state as an argument, and by adjusting its own parameters as appropriate, it adjusts the amount of lactate produced in relation to the exercise state. The function part is not particularly limited as long as it is a sigmoid function, and a standard sigmoid function or an exponential sigmoid function can be selected as appropriate.

[0055] The differences from the non-patent literature (original research paper) are that the modified Hovorka model in this embodiment derives the lactate value using formula (3), and that a term representing the exercise state (sigmoid(Ψ)) has been added. The modified Hovorka model according to this embodiment simulates glucose metabolism while also considering the lactate value, which fluctuates according to the exercise state.

[0056] The inventors of this application conducted experiments using the above-described estimation model 50, second estimation model 60, and Hovorka model, and compared the control of blood glucose levels using estimation model 50 and second estimation model 60, respectively. The data period was one week (168 hours), and eating speed, exercise status, and basal administration speed were randomly sampled using a Poisson distribution. For bolus dose (administration speed), administration was performed using random carbohydrate counting according to a normal distribution, centering on a dose of 11 U. The experimental results are shown in Table 1 below.

[0057]

[0058] Table 1 shows that "Mono" represents the second estimation model 60, which estimates blood glucose levels from glucose levels without using lactate levels. "Dual" represents the estimation model 50, which estimates blood glucose levels using both glucose and lactate levels. As shown in Table 1, the TIR (Time In Range) improved to 93% in estimation model 50 compared to 83% in the second estimation model 60. In addition, the TBR (Time Below Range, < 70 mg / dL) and VTBR (Time Very Below Range, < 54 mg / dL) decreased from 8% to 1% and from 4% to 1%, respectively, indicating that the time during which insulin is overactive due to exercise is shortened. Thus, by considering not only glucose levels but also lactate levels, blood glucose levels can be appropriately estimated and insulin dosage can be controlled. Furthermore, insulin dosage can be controlled to reflect exercise levels without the need for a separate monitoring device to monitor exercise status or for the patient to input data themselves.

[0059] Figure 6 is a flowchart showing the procedure for generating the estimation model 50. Based on Figure 6, the processing content for generating the estimation model 50 by machine learning will be explained. The control unit 11 of the server 1 acquires training data to be used to generate the estimation model 50 (step S11). The training data is time-series data of continuous glucose values, continuous lactate values, insulin dosage (basal administration rate, bolus administration rate), and patient eating rate, with the correct values ​​of the difference from the current lactate value after a predetermined time and the difference from the current blood glucose value after a predetermined time labeled.

[0060] Based on training data, the control unit 11 generates an estimation model 50 that estimates the blood glucose level after a predetermined time when time-series data such as the patient's glucose level and lactate level are input (step S12). Specifically, as described above, the control unit 11 generates an estimation model 50 consisting of a first network 51 that estimates the difference from the current value of the lactate level after a predetermined time when time-series data such as the patient's glucose level and lactate level are input, and a second network 52 that estimates the difference from the current value of the blood glucose level after a predetermined time when time-series data such as the patient's glucose level and lactate level are input. The control unit 11 then completes the series of processes.

[0061] Figure 7 is a flowchart illustrating the procedure for controlling insulin administration. Based on Figure 7, the process of estimating blood glucose levels after a predetermined time from the patient's glucose and lactate levels, and controlling (determining) the insulin dosage will be explained. The control unit 21 of terminal 2 acquires time-series data of the patient's glucose and lactate levels (step S31). Specifically, in addition to actual glucose and lactate levels, the control unit 21 acquires data such as actual and planned insulin dosages for the patient, and actual and planned eating speeds for the patient.

[0062] The control unit 21 inputs the acquired time-series data of glucose value, lactate value, insulin dosage, and eating speed into the first network 51 of the estimation model 50 to estimate the difference from the current lactate value after a predetermined time (step S32). The control unit 21 inputs the acquired time-series data of glucose value, lactate value, insulin dosage, and eating speed, along with the estimated difference from the current lactate value after a predetermined time, into the second network 52 of the estimation model 50 to estimate the difference from the current blood glucose value after a predetermined time (step S33). The control unit 21 estimates the blood glucose value after a predetermined time based on the estimated difference and the current blood glucose value (step S34).

[0063] The control unit 21 determines whether the estimated blood glucose level exceeds the first threshold (step S35). If it determines that the level exceeds the first threshold (S35: YES), the control unit 21 increases the insulin dosage (basal administration rate) by a predetermined value (step S36).

[0064] If the control unit 21 determines that the estimated blood glucose level does not exceed the first threshold (S35: NO), it determines whether the estimated blood glucose level is below the second threshold (step S37). If it determines that the blood glucose level is not below the second threshold (S37: NO), the control unit 21 terminates the series of processes.

[0065] If the estimated blood glucose level is determined to be below the second threshold (S37: YES), the control unit 21 reduces the insulin dose to a predetermined value (step S38).

[0066] After executing the process in step S36 or S38, the control unit 21 outputs a control signal to the infusion device 4 to vary (increase or decrease) the insulin dosage (step S39), and the series of processes ends.

[0067] Based on the above, this embodiment allows for the appropriate determination of the insulin dosage to be administered to the patient.

[0068] The embodiments disclosed herein should be considered in all respects to be illustrative and not restrictive. The scope of the invention is indicated by the claims and not in the sense described above, and all modifications within the sense and scope equivalent to the claims are intended to be included.

[0069] The matters described in each embodiment can be combined with each other. Furthermore, the independent and dependent claims described in the claims can be combined with each other in any combination, regardless of the form of reference. In addition, the claims use a form in which claims referencing two or more other claims (multi-claim form), but are not limited to this. A form in which multi-claims referencing at least one multi-claim (multi-multi-claim) may also be used.

[0070] 1 Server 11 Control Unit 12 Main Memory Unit 13 Communication Unit 14 Auxiliary Storage Unit P1 Program 2 Terminal 21 Control Unit 22 Main Memory Unit 23 Communication Unit 24 Display Unit 25 Input Unit 26 Auxiliary Storage Unit P2 Program 50 Estimated Model 60 Second Estimated Model 3 Sensor 4 Injection Device

Claims

1. A program that causes a computer to perform the following processes: acquire time-series data showing a patient's glucose and lactate levels over time; estimate the patient's blood glucose level after a predetermined time based on the time-series data of glucose and lactate levels; and determine the amount of insulin to administer to the patient based on the estimated blood glucose level.

2. The program according to claim 1, which estimates the blood glucose level after a predetermined time by inputting the acquired time-series data of glucose and lactate values ​​into a model that has been trained to estimate the blood glucose level after a predetermined time when time-series data of glucose and lactate values ​​are input.

3. The program according to claim 2, wherein the model comprises a first network that estimates the difference from the current value of the lactate value after a predetermined time when time-series data of the glucose value and lactate value are input, and a second network that estimates the difference from the current value of the blood glucose value after a predetermined time when time-series data of the glucose value and lactate value and the difference from the current value of the lactate value after a predetermined time are input, and the program estimates the difference from the current value of the blood glucose value after a predetermined time when the acquired time-series data of the glucose value and lactate value are input to the first network, and the program estimates the difference from the current value of the blood glucose value after a predetermined time when the acquired time-series data of the glucose value and lactate value and the estimated difference from the current value of the lactate value after a predetermined time are input to the second network.

4. The program according to claim 1, which further acquires time-series data showing the actual and planned amounts of insulin administered to the patient over time, and estimates the blood glucose level after a predetermined time based on the time-series data of glucose value, lactate value, and dosage.

5. The program according to claim 1, which further acquires time-series data showing the actual and planned values ​​of the patient's eating speed in a time series, and estimates the blood glucose level after a predetermined time based on the time-series data of glucose value, lactate value, and eating speed.

6. The program according to claim 1, wherein if the estimated blood glucose level exceeds a first threshold, the insulin dose is increased from the current value by a predetermined value, and if the estimated blood glucose level is below a second threshold, the insulin dose is decreased from the current value by a predetermined value.

7. An information processing method in which a computer performs the following processes: acquires time-series data showing a patient's glucose and lactate levels over time; estimates the patient's blood glucose level after a predetermined time based on the time-series data of glucose and lactate levels; and determines the amount of insulin to be administered to the patient based on the estimated blood glucose level.

8. An information processing device comprising a control unit, wherein the control unit acquires time-series data showing a patient's glucose and lactate values ​​in time series, estimates the patient's blood glucose level after a predetermined time based on the time-series data of glucose and lactate values, and determines the amount of insulin to be administered to the patient based on the estimated blood glucose level.

Images