Simulation model generation device, simulation model generation method, and program

Abstract

A simulation model generation device (14) temporarily determines variation factor data in a first period by allocating variation factor data during a second period (< first period) to a contribution opportunity of each variation factor in a period that is included in the first period and not included in the second period, sets the temporarily determined variation factor data in a mathematical model for obtaining time-series estimation data of a glucose value of a patient, and sets the optimal solution for a parameter, obtained by performing optimization processing on a parameter included in the mathematical model, in the mathematical model so that the estimation data obtained by means of the mathematical model approximates measurement data, thereby generating a simulation model for estimating the glucose value of the patient.

Description

Simulation model generation device, simulation model generation method, and program

[0001] The present disclosure relates to a simulation model generation device, a simulation model generation method, and a program.

[0002] Japanese Patent No. 7290624 discloses a technique for estimating a patient's blood glucose level based on the amount of food consumed by the patient and the amount of insulin administered by the patient.

[0003] Patent No. 7290624

[0004] It is desirable to have a better estimate of a patient's blood glucose level.

[0005] The present invention aims to solve the above-mentioned problems.

[0006] (1) A first aspect of the present disclosure is a simulation model generation device including: a first acquisition unit that acquires measurement data, which is time-series data of a patient's glucose level during a first period; a second acquisition unit that acquires variation factor data, which is data related to variation factors that caused a variation in the patient's glucose level during a second period that is included in the first period but is shorter than the first period; a determination unit that provisionally determines the variation factor data for the first period by allocating the variation factor data acquired by the second acquisition unit to contribution opportunities of each of the variation factors during a period that is included in the first period but not included in the second period; an optimization unit that sets the variation factor data provisionally determined by the determination unit in a mathematical model for obtaining inferred data of the time series of the patient's glucose level, and performs optimization processing on parameters included in the mathematical model so that the inferred data obtained by the mathematical model approximates the measurement data acquired by the first acquisition unit, thereby obtaining an optimal solution of the parameters; and a generation unit that generates a simulation model for inferring the patient's glucose level by setting the optimal solution of the parameters obtained by the optimization unit in the mathematical model.

[0007] In order to estimate a patient's glucose level, it is necessary to set parameters corresponding to the patient in advance for a predetermined mathematical model. Furthermore, in order to increase the accuracy of glucose value calculation, it is necessary to increase the accuracy of the parameters. To increase the accuracy of the parameters, it is desirable to identify the parameters using a sufficient amount of variable factor data (e.g., dietary data, insulin data, etc.). However, the work of acquiring a sufficient amount of variable factor data and manually inputting a sufficient amount of variable factor data into the device places a burden on the patient or user (e.g., a doctor).

[0008] In the configuration of the above item (1), the determination unit provisionally determines the variation factor data for the first period by assigning the variation factor data for a second period, which is shorter than the first period, to the contribution opportunities of each variation factor for a period (third period) that is included in the first period but not included in the second period. Therefore, the patient does not need to obtain the variation factor data for the third period. Furthermore, the patient or user (doctor, etc.) does not need to input the variation factor data for the third period into the device. The configuration of the above item (1) can reduce the burden on the patient or user (doctor, etc.). Therefore, the configuration of the above item (1) simplifies the process of estimating the patient's blood glucose level.

[0009] According to the configuration of the above item (1), an optimal solution for the parameters included in the predetermined mathematical model can be obtained. That is, according to the configuration of the above item (1), a simulation model corresponding to the patient can be generated by setting the optimal solution for the parameters in the predetermined mathematical model. By using the simulation model corresponding to the patient, a user (such as a doctor) can better estimate the patient's blood glucose level.

[0010] (2) In the simulation model generation device described in item (1) above, the variable factor data acquired by the second acquisition unit may include meal amount data indicating the amount of food consumed by the patient at each of a plurality of meal occasions during the second period, meal time data corresponding to each of the plurality of meal occasions during the second period, insulin dosage data indicating the amount of insulin administered to the patient at each of a plurality of insulin administration occasions during the second period, and insulin administration time data corresponding to each of the plurality of insulin administration occasions during the second period.

[0011] (3) The simulation model generating device according to the above item (1) or (2) may further include a distribution chart acquiring unit that generates the estimated data of the time series of the glucose level as data distribution information for each time.

[0012] (4) In the simulation model generation device described in the above item (3), the second period may be 24 hours.

[0013] (5) A second aspect of the present disclosure is a method for generating a simulation model, the method comprising: a first acquisition step of acquiring measurement data, which is time-series data of a patient's glucose level during a first period; a second acquisition step of acquiring variation factor data, which is data on variation factors that caused a variation in the patient's glucose level during a second period that is included in the first period but is shorter than the first period; a determination step of provisionally determining the variation factor data for the first period by allocating the variation factor data acquired in the second acquisition step to contribution opportunities of the variation factors for each period that is included in the first period but not included in the second period; an optimization step of setting the variation factor data provisionally determined in the determination step in a mathematical model for obtaining inferred data of the time series of the patient's glucose level, and performing an optimization process on parameters included in the mathematical model so that the inferred data obtained by the mathematical model and the measurement data acquired in the first acquisition step approximate each other, thereby obtaining an optimal solution of the parameters; and a generation step of generating a simulation model for inferring the patient's glucose level by setting the optimal solution of the parameters obtained in the optimization step in the mathematical model.

[0014] In the configuration of the above item (5), in the determination step, the variation factor data for the first period is provisionally determined by allocating variation factor data for a second period, which is shorter than the first period, to the contribution opportunities of each variation factor for a period (third period) that is included in the first period but not included in the second period. Therefore, the patient does not need to obtain variation factor data for the third period. Furthermore, the patient or user (doctor, etc.) does not need to input variation factor data for the third period. The configuration of the above item (5) can reduce the burden on the patient or user (doctor, etc.). Therefore, the configuration of the above item (5) simplifies the process of estimating the patient's blood glucose level.

[0015] According to the configuration of the above item (5), an optimal solution for the parameters included in the predetermined mathematical model can be obtained. That is, according to the configuration of the above item (5), a simulation model corresponding to the patient can be generated by setting the optimal solution for the parameters in the predetermined mathematical model. By using the simulation model corresponding to the patient, a user (such as a doctor) can better estimate the patient's blood glucose level.

[0016] (6) In the method for generating a simulation model described in item (5) above, the variable factor data acquired in the second acquisition step may include meal amount data indicating the amount of food consumed by the patient at each of a plurality of meal occasions during the second period, meal time data corresponding to each of the plurality of meal occasions during the second period, insulin dosage data indicating the amount of insulin administered to the patient at each of a plurality of insulin administration occasions during the second period, and insulin administration time data corresponding to each of the plurality of insulin administration occasions during the second period.

[0017] (7) The method for generating a simulation model according to the above item (5) or (6) may further include a distribution chart acquisition step of generating the estimated data of the time series of the glucose value as data distribution information for each time.

[0018] (8) In the method for generating a simulation model described in the above item (7), the second period may be 24 hours.

[0019] (9) A third aspect of the present disclosure is a program for causing a computer to execute the simulation model generation method described in any one of items (5) to (8) above.

[0020] The present invention allows for a better estimation of a patient's blood glucose level.

[0021] Fig. 1 is a schematic diagram of a clinical decision support system. Fig. 2 is a flowchart of a simulation model generation process. Fig. 3 is a graph showing a patient's measurement data for seven days, and the amount of food eaten and the amount of insulin administered on the first day. Fig. 4 is a graph showing a patient's measurement data for seven days, and the amount of food eaten and the amount of insulin administered on each of the first to seventh days. Fig. 5 is a graph displayed on the display unit.

[0022] In this specification, the subject of glucose measurement (glucose concentration) is referred to as a "patient." Furthermore, the operator who operates the clinical decision support device is referred to as a "user." For example, the user may be a medical professional such as a doctor, a patient, or a person providing care to the patient. Note that in this specification, the term "glucose level" simply refers to the glucose level in the interstitial fluid under the patient's skin. The glucose level in the interstitial fluid is correlated with the blood glucose level. Therefore, the glucose level in this specification may be treated as synonymous with the blood glucose level.

[0023] In order to estimate a patient's glucose level, it is necessary to set parameters corresponding to the patient in advance for a predetermined mathematical model. To improve the calculation accuracy of the estimated glucose level, it is necessary to improve the estimation accuracy of the parameters. To improve the estimation accuracy of the parameters, it is desirable to identify the parameters using a sufficient amount of data (e.g., data related to meals, data related to insulin, etc.). However, the work of acquiring a sufficient amount of data and inputting a sufficient amount of data into the device places a burden on the patient or user. The embodiment described below makes it possible to improve the calculation accuracy of the glucose level while reducing the burden on the patient or user.

[0024] 1 is a schematic diagram of the clinical decision support system 10. The clinical decision support system 10 includes a CGM (Continuous Glucose Monitoring) device 12, a clinical decision support device 14, and a server 16. The CGM device 12 and the server 16 are communicatively connected to each other via a communication line 18 such as the Internet. Similarly, the clinical decision support device 14 and the server 16 are communicatively connected to each other via the communication line 18.

[0025] Although the present specification describes a clinical decision support system 10 including a server 16, the server 16 is not an essential component of the clinical decision support system 10. For example, if the server 16 is not included in the clinical decision support system 10, the CGM device 12 and the clinical decision support device 14 are communicatively connected to each other wirelessly or via a wire. For example, the CGM device 12 and the clinical decision support device 14 may be communicatively connected to each other via a communication line 18, or may be communicatively connected to each other via short-range wireless communication such as Bluetooth (registered trademark). Alternatively, the measurement data acquired by the CGM device 12 may be compiled into a compatible file format, and the clinical decision support device 14 may read the data.

[0026] [1-1 CGM Device 12] The CGM device 12 measures the glucose level in the interstitial fluid under the patient's skin. The CGM device 12 includes a sensor device 20, which is an on-body device, and a data acquisition device 22. The sensor device 20 and the data acquisition device 22 are connected to each other so that they can communicate with each other via wire or wirelessly.

[0027] The sensor device 20 is worn on the patient's body. The sensor device 20 includes a sensor and a transmitter (neither shown). The sensor continuously measures glucose levels in the patient's subcutaneous interstitial fluid. The transmitter transmits a signal indicative of the glucose level measured by the sensor to a data acquisition device 22.

[0028] The data acquisition device 22 is carried by the patient. The data acquisition device 22 may be, for example, a mobile terminal such as a smartphone. The data acquisition device 22 includes a receiver, a controller (processor, etc.), a monitor screen, a transmitter, a memory, and an input unit (none of which are shown). The receiver receives a signal indicating the glucose value transmitted from the sensor device 20. The controller displays the patient's glucose value on the monitor screen. The transmitter transmits a signal indicating the glucose value to the server 16 via the communication line 18. The memory stores the time-series glucose values. The patient may input variable factor data into the data acquisition device 22 via the input unit.

[0029] [1-2 Clinical decision support device 14] The clinical decision support device 14 is an information processing device with a function to simulate a patient's blood glucose trend. The clinical decision support device 14 also has a function to generate a simulation model for simulating a patient's blood glucose trend. That is, the clinical decision support device 14 also functions as a simulation model generation device. After generating a simulation model corresponding to the patient, the clinical decision support device 14 simulates the patient's blood glucose trend.

[0030] The clinical decision support device (simulation model generation device) 14 generates a simulation model corresponding to a patient based on the "measurement data" and "variation factor data." The clinical decision support device 14 also obtains "estimated data" by setting the patient's "variation factor data" in the simulation model. Before explaining the configuration of the clinical decision support device 14, the names of each piece of data will be defined.

[0031] The measurement data is time-series data of the patient's glucose levels over a first period (e.g., three days, one week, one month, etc.). The measurement data is the glucose concentration in the patient's interstitial fluid measured continuously over a predetermined period. The measurement data is typically obtained once every one to five minutes. In this embodiment, the measurement data is time-series data of glucose levels measured by the CGM device 12 and is stored in the server 16.

[0032] The fluctuation factor data is data related to fluctuation factors during a second period (e.g., 24 hours) that is included in the first period but is shorter than the first period. The fluctuation factor is a patient's behavior (disturbance) that triggers fluctuations in the patient's glucose level, such as food intake and insulin administration. That is, the fluctuation factor data includes dietary data related to food intake and insulin data related to insulin administration. The dietary data includes food amount data and meal time data corresponding to the food amount data. The insulin data includes insulin dosage data and insulin administration time data corresponding to the insulin dosage data.

[0033] The meal amount data is data indicating the amount of carbohydrates or sugars ingested by the patient at each of multiple meal occasions (breakfast, lunch, dinner, snacks, etc.) during the second period (24 hours). The meal time data is data indicating the time at which each of the multiple meal occasions occurred during the second period. The insulin dosage data is data indicating the amount of insulin administered to the patient at each of multiple insulin administration occasions (before or during meals, before bedtime, etc.) during the second period. The insulin administration time data is data indicating the time at which each of the multiple insulin administration occasions occurred during the second period.

[0034] The patient can store the variable factor data in the server 16 via, for example, a communication terminal (such as the CGM device 12 or a personal computer (not shown)) owned by the patient.

[0035] The estimated data is time-series data of the patient's glucose level obtained by a simulation model, and is obtained by setting the patient's variable factor data for a given period in a simulation model corresponding to the patient.

[0036] The clinical decision support device 14 acquires the time-series data and the variable factor data from the server 16 via the communication line 18. Although this specification describes an embodiment in which the variable factor data is acquired from the server 16, the variable factor data may also be input by a user (such as a doctor) operating the input unit 30. For example, the patient may report the variable factor data to the user. In this case, the user operates the input unit 30 to input the variable factor data reported by the patient into the control unit 34 (storage unit 38).

[0037] The clinical decision support device 14 may be, for example, a terminal such as a personal computer, a tablet, etc. The clinical decision support device 14 includes an input unit 30, a display unit 32, and a control unit 34. The control unit 34 further includes a calculation unit 36 ​​and a storage unit 38.

[0038] The input unit 30 is a user-machine interface that can be operated by a user. The input unit 30 may include a mouse, a touch panel, a keyboard, a voice input device, etc. The input unit 30 transmits a signal corresponding to an input operation by the user to the control unit 34. The display unit 32 displays an image corresponding to a video signal transmitted from the control unit 34.

[0039] The calculation unit 36 ​​of the control unit 34 may be configured by a processor such as a central processing unit (CPU) or a graphics processing unit (GPU). That is, the calculation unit 36 ​​may be configured by processing circuitry. At least a part of the calculation unit 36 ​​may be realized by an integrated circuit such as an application specific integrated circuit (ASIC) or a field-programmable gate array (FPGA). At least a part of the calculation unit 36 ​​may be realized by an electronic circuit including discrete devices.

[0040] The calculation unit 36 ​​includes a simulation model generation unit 40, an acquisition unit (distribution chart acquisition unit) 42, and a display control unit 44. The simulation model generation unit 40, the acquisition unit 42, and the display control unit 44 can be realized by the calculation unit 36 ​​executing a program stored in the storage unit 38.

[0041] The simulation model generation unit 40 identifies parameters corresponding to the patient by performing an optimization process on parameters included in a predetermined mathematical model for estimating glucose levels. As a result, the simulation model generation unit 40 generates a mathematical model including parameters corresponding to the patient. The predetermined mathematical model including parameters corresponding to the patient is referred to as a simulation model corresponding to the patient. Specific examples of the predetermined mathematical model will be described later.

[0042] The simulation model generation unit 40 includes a first acquisition unit 48, a second acquisition unit 50, a determination unit 52, an optimization unit 54, and a generation unit 56. The first acquisition unit 48 acquires measurement data measured by the CGM device 12 during a first period (e.g., one week, one month, etc.) from the server 16 in response to a user's operation of the input unit 30. The second acquisition unit 50 acquires the above-mentioned variation factor data for the second period (e.g., 24 hours, etc.) from the server 16 in response to a user's operation of the input unit 30 or the data acquisition device 22 of the CGM device 12. Alternatively, the second acquisition unit 50 reads variation factor data directly input to the input unit 30.

[0043] The determination unit 52 allocates the fluctuation factor data acquired by the second acquisition unit 50 to the contribution opportunities (meal opportunities, insulin administration opportunities) of each fluctuation factor in a third period that is included in the first period but not included in the second period, thereby provisionally determining the fluctuation factor data for the third period.

[0044] The optimization unit 54 performs optimization processing to obtain an optimal solution for the variation factors for the first period (second period + third period) and the parameters included in a predetermined mathematical model. The optimization unit 54 optimizes the variation factor data for the first period provisionally determined by the determination unit 52 with respect to the measurement data for the first period. Furthermore, the optimization unit 54 optimizes the variation factor data and the parameters so that the measurement data approximates the predetermined mathematical model. By performing these processing steps, the optimization unit 54 obtains an optimal solution for the parameters included in the predetermined mathematical model. The predetermined mathematical model will be described later.

[0045] The generation unit 56 generates a simulation model for estimating the glucose level of the patient by setting the optimal solution of the parameters obtained by the optimization unit 54 in a predetermined mathematical model. The simulation model generated here is a simulation model corresponding to the patient.

[0046] The acquisition unit 42 acquires estimated data by setting the variable factor data input via the input unit 30 or the variable factor data stored in the memory unit 38 into a simulation model corresponding to the patient generated by the simulation model generation unit 40.

[0047] The display control unit 44 performs display control to display various types of information on the display unit 32. For example, the display control unit 44 performs display control to display the estimated data acquired by the acquisition unit 42 on the display unit 32. The display control unit 44 transmits a video signal to the display unit 32 to display various types of information.

[0048] The storage unit 38 is a computer-readable storage medium. The storage unit 38 is composed of a volatile memory (not shown) and a non-volatile memory (not shown). The volatile memory is, for example, a random access memory (RAM). The non-volatile memory is, for example, a read-only memory (ROM), a flash memory, etc. Data, etc. are stored in the volatile memory. Programs, tables, maps, etc. are stored in the non-volatile memory. At least a portion of the storage unit 38 may be provided in the processor, integrated circuit, etc. described above.

[0049] The storage unit 38 is a storage medium that stores programs for implementing each unit in the calculation unit 36. The storage unit 38 is also a storage medium that stores programs for executing a simulation model generation process and a clinical decision support process, which will be described below.

[0050] [1-3 Server 16] The server 16 may be provided by a physical server or a cloud server. The server 16 includes a processor and a memory (neither of which are shown).

[0051] The server 16 acquires the patient's measurement data from the data acquisition device 22 of the CGM device 12 via the communication line 18. The server 16 also acquires the patient's variable factor data from a communication terminal (such as the data acquisition device 22 or a personal computer (not shown)) owned by the patient via the communication line 18. The measurement data and variable factor data are stored in memory. The server 16 transmits a signal indicating the measurement data and variable factor data to the clinical decision support device 14 via the communication line 18.

[0052] [2 Mathematical Model] As described above, the optimization unit 54 of the simulation model generation unit 40 performs optimization processing to obtain an optimal solution for parameters included in a predetermined mathematical model. As the mathematical model, for example, the Hovorka equation disclosed in Japanese Patent No. 7290624 or the equation disclosed in the specification of U.S. Patent No. 6,923,763 can be used.

[0053] For example, the Hovorka equation is the following simultaneous ordinary differential equations with time t as a variable: In the following simultaneous ordinary differential equations, the function i(t) is the insulin dose, the function cho(t) is the meal amount (amount of carbohydrates), and the function G(t) is the glucose value.

[0054]

[0055] In the above simultaneous ordinary differential equations, F ０１ Ｃ is expressed by the following formula:

[0056] In the above simultaneous ordinary differential equations, F Ｒis 0.003(G-9) x V when G>9 Ｇ and is 0 unless G>9.

[0057] The simultaneous ordinary differential equations have 15 parameters V Ｇ , F ０１ , k １２ , F Ｒ , EGP ０ , k ｂ１ , k ａ１ , k ｂ２ , k ａ２ , k ｂ３ , k ａ３ , k ａ , V Ｉ , k ｅ , and t ｍａｘ The parameters are as follows: V Ｇ F: Volume of glucose distribution (unit: liters) ０１ : Non-insulin-dependent glucose transfer rate (unit: mmol / min) k １２ : Rate constant of glucose transfer from tissue to blood (unit: min －１ ) k ａ１ , k ａ２ , k ａ３ : Insulin inactivation rate constant (unit: min －１ ) F Ｒ : Urinary excretion constant of glucose (unit: mmol / min) EGP ０ : endogenous glucose production rate per hour (unit: min －１ ) k ｂ１ , k ｂ２ , k ｂ３ : Insulin activation rate constant (unit: min －１ ) k ａ : Absorption rate of subcutaneously injected insulin (unit: min －１ ) V Ｉ : Volume of insulin distribution (unit: liters) k ｅ : plasma elimination rate of insulin (unit: min －１ ) t ｍａｘ : Time to reach peak absorption of glucose ingested by the patient (unit: minutes)

[0058] The storage unit 38 stores a predetermined mathematical model, for example, the simultaneous ordinary differential equations.

[0059] [3 Simulation Model Generation Process] Before the simulation model generation unit 40 of the clinical decision support device 14 generates a simulation model corresponding to the patient, the CGM device 12 acquires measurement data for the first period. The server 16 acquires the measurement data for the first period from the CGM device 12. Furthermore, before the simulation model generation process, the patient inputs variable factor data for a second period, which is included in the first period but is shorter than the first period, to a communication terminal (e.g., the CGM device 12) owned by the patient. The server 16 acquires the variable factor data for the second period from the communication terminal owned by the patient. The server 16 stores the measurement data and variable factor data acquired from the communication terminal owned by the patient. Alternatively, the user can manually input the variable factor data for the second period directly into the input unit 30 of the clinical decision support device 14. In this case, it is sufficient that one day's worth of variable factor data is input as the variable factor data for the second period. Note that the variable factor data may be input by the patient or by another user, such as a medical professional.

[0060] Instead of the food intake data in the variation factor data, the patient may input information indicating specific meal contents (e.g., the name of a meal menu, actual intake amount) into the communication terminal (e.g., CGM device 12). To accommodate such cases, the memory of the communication terminal or the memory of the server 16 may pre-store a conversion table that converts information indicating meal contents into food intake data. Alternatively, the communication terminal may obtain information for converting information indicating meal contents into food intake data from a source other than the server 16. In this way, when information indicating meal contents is input, the communication terminal or server 16 can obtain food intake data based on the conversion table.

[0061] 2 is a flowchart of the simulation model generation process. The simulation model generation process is executed by the clinical decision support device (simulation model generation device) 14. A user (such as a doctor) performs a predetermined operation on the input unit 30. As a result, the input unit 30 instructs the calculation unit 36 ​​of the control unit 34 to execute the simulation model generation process. The calculation unit 36 ​​executes the simulation model generation process shown in FIG. 2 in response to the instruction signal output from the input unit 30.

[0062] In step S1, the first acquisition unit 48 acquires the patient's measurement data for a first period from the server 16 via the communication line 18. Alternatively, the first acquisition unit 48 may acquire the patient's measurement data directly from the CGM device 12 (or via the data acquisition device 22). For example, a curve 60 in Fig. 3 shows seven days' worth of measurement data for a patient.

[0063] In step S2, the second acquisition unit 50 acquires the patient's variable factor data for the second period from the server 16 via the communication line 18. The second acquisition unit 50 may acquire the variable factor data input via the input unit 30. Three dots 62 in FIG. 3 indicate dietary data (meal amount data and meal time data) for the first day (24 hours) of the first period (e.g., seven days). Four dots 64 in FIG. 3 indicate insulin data (insulin dose data and insulin administration time data) for the first day of the first period. As such, the second period is shorter than the first period. The variable factor data may be acquired from information stored on the server 16. The user may directly input representative values ​​for the first day of the first period via the input unit 30. For example, if the variable factor data is dietary data, representative values ​​of the carbohydrate amount (sugar amount) and meal time data that the patient recognizes as having ingested are input based on the patient's self-report. Alternatively, if the variable factor data is insulin data related to insulin administration, insulin dose data and insulin administration time data are directly input based on a prescription plan. The variable factor data is not limited to dietary data and insulin data, and may further include information such as the type and amount of exercise, and fever.

[0064] In step S3, the determination unit 52 provisionally determines the variation factor data for the first period. As described above, a period that is included in the first period but not included in the second period is referred to as the third period. For example, the determination unit 52 provisionally inputs the meal data (meal amount data and meal time data) for each meal occasion on the first day (second period) acquired in step S2 as the initial values ​​of the meal data for each meal occasion on each of the second to seventh days (third period). That is, the determination unit 52 also assigns the meal data for the breakfast occasion on the first day to the breakfast occasion on each of the second to seventh days. The determination unit 52 also assigns the meal data for the lunch occasion on the first day to the lunch occasion on each of the second to seventh days. The determination unit 52 also assigns the meal data for the dinner occasion on the first day to the dinner occasion on each of the second to seventh days. For snacks other than breakfast, lunch, and dinner, the determination unit 52 also allocates the meal data of the snack occasion on the first day to the snack occasions on each of the second to seventh days.

[0065] Similarly, the decision unit 52 also uses the insulin data (insulin dosage data and insulin administration time data) of the insulin administration occasion on the first day (second period) acquired in step S2 as the insulin data of each insulin administration occasion on each of the second to seventh days (third period). That is, the decision unit 52 also assigns the insulin data of the insulin administration occasion at breakfast on the first day to the insulin administration occasion at breakfast on each of the second to seventh days. The decision unit 52 also assigns the insulin data of the insulin administration occasion at lunch on the first day to the insulin administration occasion at lunch on each of the second to seventh days. The decision unit 52 also assigns the insulin data of the insulin administration occasion at dinner on the first day to the insulin administration occasion at dinner on each of the second to seventh days. The decision unit 52 also assigns the insulin data of the insulin administration occasion before going to bed on the first day to the insulin administration occasion before going to bed on each of the second to seventh days.

[0066] By the process of step S3 described above, the fluctuation factor data for each day of the seven days (first period) is provisionally determined as shown in FIG.

[0067] In step S4, the optimization unit 54 performs a process (optimization process) to optimize the parameters and fluctuation factor data included in a predetermined mathematical model (e.g., the simultaneous ordinary differential equations). As a result, the optimization unit 54 acquires parameters corresponding to the patient's measurement data acquired by the first acquisition unit 48 in step S1. These parameters can be treated as parameters corresponding to the patient. At the end of step S3, the parameters included in the predetermined mathematical model are unknown. If parameters corresponding to the patient are set in the predetermined mathematical model, the time-series inferred data obtained by setting the fluctuation factor data for the first period in the predetermined mathematical model should approximate the measurement data acquired in step S1. Therefore, the optimization unit 54 sets the fluctuation factor data for the first period acquired in step S3 in a predetermined mathematical model including unknown parameters. Then, the optimization unit 54 performs an optimization process for the parameters of the mathematical model so that the time-series inferred data obtained by the predetermined mathematical model approximates the measurement data acquired in step S1. The optimization unit 54 performs an optimization process using at least one of predetermined optimization methods (e.g., steepest gradient method, quasi-Newton method, Newton method, Markov chain-Monte Carlo method, Bayesian optimization, etc.). As a result, the optimization unit 54 obtains an optimal solution for the parameters included in the predetermined mathematical model. Furthermore, the optimization unit 54 performs an optimization process using at least one of the predetermined optimization methods for each data included in the variation factor data to obtain an optimal solution for the variation factor data provisionally determined in step S3. Steps S3 and S4 may be performed multiple times. As a result, the optimization unit 54 fine-tunes the provisionally determined variation factor data.

[0068] In step S5, the generator 56 sets the optimal solution of the parameters obtained by the optimizer 54 in a predetermined mathematical model. As a result, the generator 56 generates a simulation model for estimating the patient's glucose level. The generator 56 stores the generated simulation model and the fine-tuned fluctuation factor data obtained by the optimizer 54 in the memory 38.

[0069] In step S6, the acquisition unit 42 estimates time-series data of glucose levels by setting the fluctuation factor data in a simulation model. Specifically, the acquisition unit 42 acquires estimated data by setting the fluctuation factor data determined in step S4 (the fluctuation factor data stored in the memory unit 38) in the simulation model generated in step S5. The estimated data may be generated over the same period as the first period. The acquisition unit 42 may analyze the estimated data using statistical techniques and acquire the data obtained by the analysis in any display format. In this case, for example, the variation or bias of blood glucose level fluctuations over 24 hours can be expressed using known distribution representation methods (e.g., box-and-whisker plots, percentile curves, etc. at any time). When expressed using a box-and-whisker plot, outliers may also be displayed. Hereinafter, the analysis results showing the 24-hour blood glucose level distribution obtained by analyzing the estimated data are referred to as a data distribution chart. The acquisition unit 42 may convert the estimated data into blood glucose level data.

[0070] In step S7, the display control unit 44 causes the display unit 32 to display the estimation result in step S6. Specifically, the display control unit 44 transmits a video signal to the display unit 32 to display the data distribution chart acquired in step S6 and the fluctuation factor data determined in step S4. As a result, the display unit 32 displays the data distribution chart and the fluctuation factor data. For example, the display unit 32 displays the graph shown in FIG. 5. This graph is information corresponding to the estimation data and the fluctuation factor data. This graph represents blood glucose fluctuations over 24 hours.

[0071] Fig. 5 shows a graph displayed by the display unit 32. The data distribution chart shown in Fig. 5 shows the data distribution for each time period. Each of the three points 72 shown in Fig. 5 represents meal data (meal amount data and meal time data) stored in the memory unit 38. Each of the four points 74 in Fig. 5 represents insulin data (insulin dose data and insulin administration time data) stored in the memory unit 38. In this case, the initial values ​​of the meal data and the initial values ​​of the insulin data may be the variable factor data reported by the patient or values ​​after optimization processing.

[0072] As shown in FIG. 5 , displaying insulin data and meal data, including time information, within the display area of ​​a data distribution chart showing 24-hour blood glucose fluctuations makes it easier to visually grasp the relationship between daily treatment and blood glucose trends. Note that points 72 and 74 do not need to be displayed within the graph frame. Also, display fields showing meal amount data, meal time data, insulin dose data, and insulin administration time data may be set separately near the graph. Note that while three meal data and four insulin data per day have been described here, one or more meal data may be used. The number of meal data may be equivalent to the total number of meals and snacks per day. The number of insulin administrations may also be any number.

[0073] In the above-described simulation model generation process, steps S6 and S7 are processes for the user to confirm the estimated data obtained when the optimized variable factor data is set in the simulation model.

[0074] As described above, in the clinical decision support device 14, the determination unit 52 provisionally determines the variation factor data for the first period by allocating the variation factor data for the second period, which is shorter than the first period, to the contribution opportunities of each variation factor for the third period, which is included in the first period but not included in the second period. Therefore, the patient does not need to separately obtain the variation factor data for the third period. Furthermore, the patient or user (e.g., a doctor) does not need to input the variation factor data for the third period into the device. The clinical decision support device 14 can reduce the user's effort in recording meal amounts and meal times over multiple days and in inputting meal amounts and meal times into a recording device such as the data acquisition device 22. Therefore, the clinical decision support device 14 simplifies the process of estimating a patient's blood glucose level.

[0075] The clinical decision support device 14 can obtain optimal solutions for parameters included in a predetermined mathematical model. In other words, the clinical decision support device 14 can generate a simulation model tailored to a patient by setting optimal parameter solutions in a predetermined mathematical model. By using a simulation model tailored to a patient, a user can more accurately estimate the patient's blood glucose level. More specifically, for the variable factor data in step S2, the user can arbitrarily input variable factor data for the first day of the first period via the input unit 30. Accurate time-series estimated data based on the input values ​​can be obtained each time. Furthermore, the clinical decision support device 14 can display a data distribution chart based on the estimated data. This allows the patient to view a data distribution chart estimated based on specific dietary data and insulin data, allowing them to simulate meal amounts, meal times, insulin doses, and insulin administration times, thereby encouraging them to accept and engage in treatment. In this way, a highly accurate simulation model can be obtained based on each patient's actual measurement data.

[0076] The clinical decision support device 14 can use the generated simulation model to simulate the patient's blood glucose trend.

[0077] Although the present disclosure has been described in detail, the present disclosure is not limited to the individual embodiments described above. Various additions, substitutions, modifications, partial deletions, etc. are possible in these embodiments without departing from the gist of the present disclosure or the spirit of the present disclosure derived from the content of the claims and their equivalents. These embodiments can also be implemented in combination. For example, in the above-described embodiments, the order of each operation and the order of each process are shown as examples and are not limited to these. The same applies when numerical values ​​or mathematical expressions are used in the description of the above-described embodiments.

Claims

1. A simulation model generation device comprising: a first acquisition unit that acquires measurement data, which is time-series data of a patient's glucose level during a first period; a second acquisition unit that acquires variation factor data, which is data related to variation factors that caused variations in the patient's glucose level during a second period that is included in the first period but is shorter than the first period; a determination unit that provisionally determines the variation factor data for the first period by allocating the variation factor data acquired by the second acquisition unit to contribution opportunities of each of the variation factors during periods that are included in the first period but not included in the second period; an optimization unit that sets the variation factor data provisionally determined by the determination unit in a mathematical model for obtaining prediction data of the time series of the patient's glucose level, and performs optimization processing on parameters included in the mathematical model so that the prediction data obtained by the mathematical model approximates the measurement data acquired by the first acquisition unit, thereby obtaining optimal solutions of the parameters; and a generation unit that generates a simulation model for predicting the patient's glucose level by setting the optimal solution of the parameters obtained by the optimization unit in the mathematical model.

2. A simulation model generating device as described in claim 1, wherein the variable factor data acquired by the second acquisition unit includes: meal amount data indicating the amount of food ingested by the patient at each of a plurality of meal occasions during the second period; meal time data corresponding to each of the plurality of meal occasions during the second period; insulin dosage data indicating the amount of insulin administered to the patient at each of a plurality of insulin administration occasions during the second period; and insulin administration time data corresponding to each of the plurality of insulin administration occasions during the second period.

3. A simulation model generating device according to claim 1 or 2, further comprising a distribution chart acquiring unit that generates the estimated data of the time series of glucose values ​​as data distribution information for each time.

4. A simulation model generating device according to claim 3, wherein the second period is 24 hours.

5. A method for generating a simulation model, comprising: a first acquisition step of acquiring measurement data, which is time-series data of the patient's glucose level during a first period; a second acquisition step of acquiring variation factor data, which is data related to variation factors that have caused variations in the patient's glucose level during a second period that is included in the first period but is shorter than the first period; a determination step of provisionally determining the variation factor data for the first period by allocating the variation factor data acquired in the second acquisition step to contribution opportunities of each of the variation factors for periods that are included in the first period but not included in the second period; an optimization step of setting the variation factor data provisionally determined in the determination step in a mathematical model for obtaining prediction data of the patient's time series of glucose levels, and performing optimization processing on parameters included in the mathematical model so that the prediction data obtained by the mathematical model approximates the measurement data acquired in the first acquisition step, to obtain optimal solutions of the parameters; and a generation step of generating a simulation model for predicting the patient's glucose level by setting the optimal solution of the parameters obtained in the optimization step in the mathematical model.

6. A method for generating a simulation model as described in claim 5, wherein the variable factor data acquired in the second acquisition step includes: meal amount data indicating the amount of food ingested by the patient at each of a plurality of meal occasions during the second period; meal time data corresponding to each of the plurality of meal occasions during the second period; insulin dosage data indicating the amount of insulin administered to the patient at each of a plurality of insulin administration occasions during the second period; and insulin administration time data corresponding to each of the plurality of insulin administration occasions during the second period.

7. A method for generating a simulation model according to claim 5 or 6, further comprising a distribution chart acquisition step for generating the estimated data of the time series of glucose values ​​as data distribution information for each time point.

8. A method for generating a simulation model according to claim 7, wherein the second period is 24 hours.

9. A program for causing a computer to execute the simulation model generation method according to claim 5.

Images