초록 ▼

An augmented, adaptive algorithm utilizing model predictive control (MPC) is developed for closed-loop glucose control in type 1 diabetes. A linear empirical input-output subject model is used with an MPC algorithm to regulate blood glucose online, where the subject model is recursively adapted, and

An augmented, adaptive algorithm utilizing model predictive control (MPC) is developed for closed-loop glucose control in type 1 diabetes. A linear empirical input-output subject model is used with an MPC algorithm to regulate blood glucose online, where the subject model is recursively adapted, and the control signal for delivery of insulin and a counter-regulatory agent such as glucagon is based solely on online glucose concentration measurements. The MPC signal is synthesized by optimizing an augmented objective function that minimizes local insulin accumulation in the subcutaneous depot and control signal aggressiveness, while simultaneously regulating glucose concentration to a preset reference set point. The mathematical formulation governing the subcutaneous accumulation of administered insulin is derived based on nominal temporal values pertaining to the pharmacokinetics (timecourse of activity) of insulin in human, in terms of its absorption rate, peak absorption time, and overall time of action.

대표청구항 ▼

What is claimed is: 1. A system for automatic control of blood glucose level of a subject, comprising: a glucose sensor operative to continually sense a glucose level of the subject and generate a corresponding glucose level signal; a delivery device operative to deliver doses of insulin to the sub

What is claimed is: 1. A system for automatic control of blood glucose level of a subject, comprising: a glucose sensor operative to continually sense a glucose level of the subject and generate a corresponding glucose level signal; a delivery device operative to deliver doses of insulin to the subject in response to an insulin dose control signal; and a controller operative to generate the insulin dose control signal as a function of the weight of the subject and time-varying glucose levels of the subject as represented by the glucose level signal over time, the controller employing a control algorithm including: generating the insulin dose control signal based on (a) the glucose level signal, and (b) accumulation of insulin in the subject due to finite rate of utilization, wherein the controller employs a model-predictive control algorithm by which the insulin dose control signal is generated as a value optimizing an objective function with objectives of (a) a weighted integration of a difference between a predicted glucose level signal and a setpoint signal over a time horizon, and (b) a weighted integration of the insulin dose control signal over the time horizon. 2. A system according to claim 1, wherein the model-predictive control algorithm is an adaptive model-predictive control algorithm which includes recursively and continually updating model parameters to dynamically adapt the subject model to variations in the response of the subject to the delivered doses of insulin. 3. A system according to claim 1, wherein the objective function is augmented to also have an objective of (c) minimizing the accumulation of insulin in the subject due to finite rate of utilization. 4. A system according to claim 1, wherein: (1) the control algorithm includes a subject model to explicitly model response of the subject to delivered doses of insulin and thereby generate, based on time-varying values of the glucose level signal and the insulin dose control signal, the predicted glucose level signal representing a predicted glucose level of the subject; and (2) the insulin dose control signal is generated based on a difference between the predicted glucose level signal and the setpoint signal representing a desired glucose level of the subject. 5. A system according to claim 4 wherein the subject model is an empirical subject model. 6. A system according to claim 5, wherein the empirical subject model is of a type initially constructed based on a system identification process performed on input-output data obtained from open-loop glycemic control of the subject. 7. A system according to claim 1 wherein the controller is further operative to generate the insulin dose control signal to provide a basal rate of delivery of insulin when the control algorithm reveals no need for a dose of insulin that exceeds a basal rate of delivery. 8. A system according to claim 7 wherein the basal rate is determined by a user-provided basal-rate value. 9. A system according to claim 8 wherein the controller employs a default basal-rate value based on the weight of the subject when the user-provided basal-rate value is either absent or greater than a predetermined maximum allowable value. 10. A system according to claim 9 wherein the controller adapts the basal rate online based on the glucose level signal over time. 11. A system according to claim 10 wherein the controller is operative to automatically impose respective constraints on the insulin dose control signal corresponding to a maximum allowable dose of insulin. 12. A system according to claim 1 wherein the glucose sensor is integrated with the delivery device. 13. A system according to claim 12 wherein the delivery device comprises a mechanically driven infusion mechanism and a cartridge for insulin. 14. A system according to claim 1, wherein: the delivery device is further operative to deliver doses of a counter-regulatory agent to the subject in response to a counter-regulatory-agent dose control signal; and the controller is further operative to generate the counter-regulatory-agent dose control signal as a function of the weight of the subject and the time-varying glucose levels of the subject as represented by the glucose level signal over time. 15. A system according to claim 14 wherein the counter-regulatory agent comprises glucagon. 16. A system according to claim 14 wherein generating the counter-regulatory-agent dose control signal accounts for accumulation in the subject from past doses of the counter-regulatory agent. 17. A system according to claim 1, wherein delivery by the delivery device is to a subcutaneous space in the subject. 18. A system according to claim 1 wherein the generating of the insulin dose control signal based on the accumulation of insulin in the subject provides a controlling effect on aggressiveness of the control algorithm. 19. A system according to claim 1 wherein generating the insulin dose control signal is performed based on a combination of one or more first mathematical expressions of a relationship between the insulin dose control signal and the glucose level signal and one or more second mathematical expressions of an estimation of the accumulation of insulin in the subject based on the insulin dose control signal. 20. A system according to claim 19 wherein generating the insulin dose control signal based on the glucose level signal includes scaled summations of past and ongoing components of the insulin dose control signal and the glucose level signal, and wherein the estimation of the accumulation of insulin in the subject is obtained based on a scaled summation of accumulation contributions based on the insulin dose control signal, including past, ongoing, and/or future components of the insulin dose control signal. 21. A system according to claim 19 wherein the estimation of the accumulation of insulin in the subject is obtained by feedback of the insulin dose control signal. 22. A system according to claim 19 wherein the first and second mathematical expressions are used to obtain a control objective. 23. A method for automatic control of the blood glucose level of a subject, comprising: continually sensing a glucose level of the subject and generating a corresponding glucose level signal; operating a delivery device to deliver doses of insulin to the subject in response to an insulin dose control signal; and generating the insulin dose control signal as a function of the weight of the subject and time-varying glucose levels of the subject as represented by the glucose level signal over time, by a control algorithm including: generating the insulin dose control signal based on (a) the glucose level signal, and (b) accumulation of insulin in the subject due to finite rate of utilization, wherein the control algorithm includes a model-predictive control algorithm by which the insulin dose control signal is generated as a value optimizing an objective function with objectives of (a) a weighted integration of a difference between a predicted glucose level signal and a setpoint signal over a time horizon, and (b) a weighted integration of the insulin dose control signal over the time horizon. 24. A method according to claim 23, wherein the model-predictive control algorithm is an adaptive model-predictive control algorithm which includes recursively and continually updating model parameters to dynamically adapt the subject model to variations in the response of the subject to the delivered doses of insulin. 25. A method according to claim 23, wherein the objective function is augmented to also have an objective of (c) minimizing the accumulation of insulin in the subject due to finite rate of utilization. 26. A method according to claim 23, wherein: (1) the control algorithm includes utilizing a subject model to explicitly model response of the subject to delivered doses of insulin and thereby generate, based on time-varying values of the glucose level signal and the insulin dose control signal, the predicted glucose level signal representing a predicted glucose level of the subject; and (2) the insulin dose control signal is generated based on a difference between the predicted glucose level signal and the setpoint signal representing a desired glucose level of the subject. 27. A method according to claim 26 wherein the subject model is an empirical subject model. 28. A method according to claim 27, wherein the empirical subject model is initially constructed based on a system identification process performed on input-output data obtained from open-loop glycemic control of the subject. 29. A method according to claim 23 further comprising generating the insulin dose control signal to provide a basal rate of delivery of insulin when the control algorithm reveals no need for a dose of insulin that exceeds a basal rate of delivery. 30. A method according to claim 29 wherein the basal rate is determined by a user-provided basal-rate value. 31. A method according to claim 30 wherein a default basal-rate value based on the weight of the subject is employed when the user-provided basal-rate value is either absent or greater than a predetermined maximum allowable value. 32. A method according to claim 31 wherein the basal rate is adapted online based on the glucose level signal over time. 33. A method according to claim 32 further comprising automatically imposing constraints on the insulin dose control signal corresponding to a maximum allowable dose of insulin. 34. A method according to claim 23, further comprising: operating the delivery device to deliver doses of a counter-regulatory agent to the subject in response to a counter-regulatory-agent dose control signal; and generating the counter-regulatory-agent dose control signal as a function of the weight of the subject and the time-varying glucose levels of the subject as represented by the glucose level signal over time. 35. A method according to claim 34 wherein the counter-regulatory agent comprises glucagon. 36. A method according to claim 34 wherein generating the counter-regulatory-agent dose control signal accounts for accumulation in the subject from past doses of the counter-regulatory agent. 37. A method according to claim 23, wherein delivery by the delivery device is to a subcutaneous space in the subject. 38. A method according to claim 23 wherein the generating of the insulin dose control signal based on (b) accumulation of insulin in the subject provides a controlling effect on aggressiveness of the control algorithm. 39. A method according to claim 23 wherein generating the insulin dose control signal is performed based on a combination of one or more first mathematical expressions of a relationship between the insulin dose control signal and the glucose level signal and one or more second mathematical expressions of an estimation of the accumulation of insulin in the subject based on the insulin dose control signal. 40. A method according to claim 39 wherein the relationship between the insulin dose control signal and the glucose level signal includes scaled summations of past and ongoing components of the insulin dose control signal and the glucose level signal, and wherein the estimation of the accumulation of insulin in the subject is obtained based on a scaled summation of accumulation contributions based on the insulin dose control signal, including past, ongoing, and/or future components of the insulin dose control signal. 41. A method according to claim 39 wherein the estimation of the accumulation of insulin in the subject is obtained by feedback of the insulin dose control signal. 42. A method according to claim 39 wherein the first and second mathematical expressions are used to obtain a control objective. 43. A controller for use in a system for automatic control of blood glucose level of a subject, the controller being operative to generate an insulin dose control signal as a function of the weight of the subject and time-varying glucose levels of the subject as represented by an glucose level signal over time, the glucose level signal being generated by a glucose sensor operative to continually sense a glucose level of the subject, the insulin dose control signal controlling the delivery of doses of insulin to the subject by a delivery device, the controller employing a control algorithm including: generating the insulin dose control signal based on (a) the glucose level signal, and (b) accumulation of insulin in the subject due to finite rate of utilization, wherein the control algorithm includes a model-predictive control algorithm by which the insulin dose control signal is generated as a value optimizing an objective function with objectives of (a) a weighted integration of a difference between a predicted glucose level signal and a setpoint signal over a time horizon, and (b) a weighted integration of the insulin dose control signal over the time horizon. 44. A controller according to claim 43, wherein the model-predictive control algorithm is an adaptive model-predictive control algorithm which includes recursively and continually updating model parameters to dynamically adapt the subject model to variations in the response of the subject to the delivered doses of insulin. 45. A controller according to claim 43, wherein the objective function is augmented to also have an objective of (c) minimizing the accumulation of insulin in the subject due to finite rate of utilization.