Method and apparatus for real-time estimation and control of physiological parameters

• US 20040034295A1
• Filed: 03/18/2003
• Published: 02/19/2004
• Est. Priority Date: 09/26/2000
• Status: Active Grant

• Current Assignee: AdVantage Industries LLC (Advantage Industries)
• Original Assignee: AdVantage Industries LLC (Advantage Industries)
• Inventors: Salganicoff, Marcos
• Granted Patent: US 7,460,130 B2
• Time in Patent Office: Days
• US Class Current: 600/365
• CPC Class Codes: A61B 5/14532 for measuring glucose, e.g. by tissue impedance measurement
• CPC Class Codes: A61B 5/1495 Calibrating or testing of in-vivo probes
• CPC Class Codes: G06F 16/40 of multimedia data, e.g. slideshows comprising image and additional audio data retrieval of still image data G06F16/50; retrieval of audio data G06F16/60; retrieval of video data G06F16/70
• CPC Class Codes: G06F 16/48 Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually

Abstract

A real-time glucose estimator uses a linearized Kalman filter to determine a best estimate of glucose level in real time. The real-time glucose estimator receives at least one measurement corresponding to glucose level. The measurement can be obtained with one or more sensors and is provided to the linearized Kalman filter in real time. The linearized Kalman filter has dynamic models and executes a recursive routine to determine the best estimate of glucose level based upon the measurement. Additional information can be provided to the linearized Kalman filter for initialization, configuration, and the like. Outputs of the linearized Kalman filter can be provided to a patient health monitor for display or for statistical testing to determine status of the real-time glucose estimator. The real-time glucose estimator can be implemented using a software algorithm. A real-time controller operating as an artificial pancreas uses a Kalman control algorithm to control glucose level of a patient in real time. The real-time controller receives an estimate of the patien: glucose level and a reference glucose level. The estimate of the patient glucose level can be provided by an optimal estimator implemented using a linearized Kalman filter. The estimated glucose level and the reference glucose level are processed by the Kalman control algorithm to determine a control command in real time. The Kalman control algorithm has a dynamic process forced by the control command a cost function determining a relative level of control. The control command is provided to a dispenser which secretes insulin or glucagon in response to the control command to correct a relatively high glucose level or a relatively low glucose level.

Claims

1. A method for providing a best estimate of glucose level in real time comprising the acts of:

obtaining a measurement which is a function of glucose level, wherein noise associated with the measurement is within limits of a predefined measurement uncertainty;

supplying the measurement to an extended Kalman filter in real time, wherein the extended Kalman filter has a dynamic process model, a dynamic measurement model, a state vector with at least one element corresponding to glucose level, and an error covariance matrix of the state vector; and

determining the best estimate of glucose level in real time using the extended Kalman filter.

2. The method of claim 1, wherein the extended Kalman filter is implemented using a software algorithm.

3. The method of claim 1, wherein determination of the best estimate of glucose level in real time uses a recursive process comprising the acts of:

computing a current estimate of the state vector using a preceding best estimate of the state vector and the dynamic process model;

computing a current error covariance matrix of the state vector using a preceding error covariance matrix of the state vector, uncertainties associated with the dynamic process model, and the dynamic process model linearized about the current estimate of the state vector;

computing a Kalman gain using the current covariance matrix, uncertainties associated with the dynamic measurement model, and the dynamic measurement model linearized about the current estimate of the state vector;

computing a new error covariance matrix of the state vector using the current error covariance matrix of the state vector, the Kalman gain, and the dynamic measurement model linearized about the current estimate of the state vector; and

computing a new best estimate of the state vector using the current estimate of the state vector, the Kalman gain, the measurement, and the dynamic measurement model.

4. The method of claim 1, wherein development of the dynamic process model and the dynamic measurement model is an iterative process comprising the acts of:

defining estimation variables and uncertain parameters which become elements of the state vector,

defining nominal time propagation of the estimation variables and the uncertain parameters;

defining a nominal sensor measurement model;

defining relationships between the estimation variables, the uncertain parameters, and the measurement;

defining uncertainties associated with the estimation variables, the uncertain parameters, and the measurement; and

verifying a nominal dynamic process model and a nominal measurement model.

5. The method of claim 4, wherein a database of measurements is used to empirically verify the nominal dynamic process model and the nominal measurement model.

6. The method of claim 5, wherein the database has a plurality of sensor measurements with corresponding direct measurements.

7. A realtime glucose estimator comprising:

a plurality of measurement inputs, wherein at least one of the measurement inputs is configured to receive an input indicative of glucose level;

a plurality of control inputs; and

an extended Kalman filter algorithm configured to receive the plurality of measurement inputs and the plurality of control inputs to provide an optimal estimate of glucose level in real time.

8. The real-time glucose estimator of claim 7, wherein at least two independent glucose sensors are coupled to the measurement inputs respectively.

9. The real-time glucose estimator of claim 8, wherein at least two independent glucose sensors are of different types.

10. The real-time glucose estimator of claim 7, wherein at least one of the measurement inputs is configured to receive a direct measurement of glucose.

11. The real-time glucose estimator of claim 10, wherein the direct measurement is derived from a capillary blood glucose measurement.

12. The real-time glucose estimator of claim 7, wherein at least one of the control inputs is configured to receive configuration data.

13. The real-time glucose estimator of claim 7, wherein at least one of the control inputs is configured to receive environment data.

14. The real-time glucose estimator of claim 7, wherein the extended Kalman filter algorithm produces the optimal estimate of glucose level in real time using a recursive loop formed by a time update module and a measurement update module.

15. The real-time glucose estimator of claim 14, wherein the plurality of control inputs is processed in real time by the time update module.

16. The real-time glucose estimator of claim 14, wherein the plurality of measurement inputs is processed in real time by the measurement update module.

17. The real-time glucose estimator of claim 16, wherein measurement inputs of a substantially identical time interval are processed serially in one cycle of the measurement update module.

18. The real-time glucose estimator of claim 10, wherein the direct measurement is provided during initialization of the real-time glucose estimator and intermittently thereafter.

19. The realtime glucose estimator of claim 7, wherein at least one of the measurement inputs is provided periodically to the realtime glucose estimator.

20. The real-time glucose estimator of claim 7, wherein at least one of the measurement inputs is coupled to a noninvasive glucose sensor.

21. The real-time glucose estimator of claim 7 further comprising a patient health monitor configured to accept user inputs, to derive configuration data based on the user inputs, and to provide the configuration data in real time to the realtime glucose estimator.

22. The realtime glucose estimator of claim 7 further comprising a patient health monitor configured to display in real time the optimal estimate of glucose level and to provide an audible alarm when the optimal estimate of glucose level is outside a predetermined range.

23. The realtime glucose estimator of claim 10 further comprising a patient health monitor configured to display in real time a performance status for the real-time glucose estimator, wherein the performance status is based upon results of statistical testing performed on residuals of the extended Kalman filter.

24. The real-time glucose estimator of claim 23, wherein the patient health monitor outputs a request for the direct measurement of glucose when the performance status indicates the optimal estimate of glucose level is outside predetermined limits.

25. An estimator for monitoring a physiological parameter comprising:

a sensor which outputs a measurement as a function of the physiological parameter;

an electronic processor coupled to an output of the sensor, wherein the electronic processor executes an algorithm that implements an extended Kalman filter to estimate the physiological parameter in real time; and

an interface coupled to an output of the electronic processor to display the estimate of the physiological parameter in real time.

26. The estimator of claim 25, wherein the interface accepts user inputs and provides a control signal in real time to the electronic processor based on the user inputs.

27. The estimator of claim 26, wherein the user inputs describe environmental conditions of the sensor in real time.

28. The estimator of claim 26, wherein the user inputs describe activities of a patient using the estimator in real time.

29. The estimator of claim 26, wherein the user inputs describe realtime administration of medication to a patient using the estimator.

30. The estimator of claim 25, wherein the estimator is a portable device contained in a relatively small package and battery operated.

31. An estimator comprising:

means for obtaining a measurement related to a physiological parameter; and

means for processing the measurement in real time using a linearized Kalman filter algorithm to provide a realtime estimate of the physiological parameter.

32. The estimator of claim 31 further comprising means for displaying the realtime estimate of the physiological parameter in real time.

33. The estimator of claim 31 further comprising means for updating the linearized Kalman filter algorithm in real time with user inputs.

34. The estimator of claim 31 further comprising means for indicating performance status of the estimator in real time.

35. A method for controlling glucose level in real time comprising the acts of:

receiving an estimated glucose level;

receiving a reference signal indicative of a desired glucose level;

providing the estimated glucose level and the reference signal to a Kalman control algorithm in real time;

determining a control command in real time using the Kalman control algorithm; and

providing the control command to a dispenser which outputs medication in response to the control command.

36. The method of claim 35, wherein the estimated glucose level is provided by an optimal estimator implemented using an extended Kalman filter.

37. The method of claim 35, wherein the reference signal is provided by a patient health monitor which accepts inputs from a user.

38. The method of claim 35, wherein the reference signal varies with time.

39. The method of claim 35, wherein the Kalman control algorithm has a dynamic process model forced by the control command and a cost function determining a relative level of control.

40. The method of claim 39, wherein determination of the control command in real time comprises the acts of:

computing a Kalman control gain to minimize the cost function; and

adjusting the control command based on the Kalman control gain and a difference between the estimated glucose level and the desired glucose level.

41. The method of claim 35, wherein the dispenser secretes insulin or glucagon in response to the control command to correct a relatively high or a relatively low estimated glucose level.

42. A method for close-loop control of a physiological parameter comprising the acts of:

obtaining a measurement of the physiological parameter from a patient;

providing the measurement to an optimal estimator in real time, wherein the optimal estimator outputs a best estimate of the physiological parameter in real time based on the measurements;

providing the best estimate of the physiological parameter to an optimal controller in real time, wherein the optimal controller outputs a control command in real time based on the best estimate of the physiological parameter and a control reference; and

providing the control command to an actuator, wherein the actuator provides an output to adjust the physiological parameter.

43. The method of claim 42, wherein the measurement is obtained using a sensor.

44. The method of claim 42, wherein the optimal estimator is implemented using a linearized Kalman algorithm.

45. The method of claim 44, wherein the optimal controller is implemented using a Kalman control algorithm.

46. The method of claim 45, wherein the optimal estimator and the optimal controller have substantially identical dynamic process models forced by the control command.

47. The method of claim 46, wherein the optimal estimator provides a best estimate state vector to the optimal controller, and the best estimate of the physiological parameter is an element of the best estimate state vector.

48. The method of claim 46, wherein the optimal controller provides the control command to the optimal estimator.

49. The method of claim 42 wherein the optimal estimator and the optimal controller are implemented as a joint Kalman algorithm.

50. The method of claim 42, wherein the control reference is provided by a patient health monitor.

51. The method of claim 42, wherein the optimal estimator outputs best estimates of additional physiological parameters to the optimal controller in real time, and the optimal controller controls the additional physiological parameters by outputting additional control commands.

52. A realtime optimal glucose controller comprising:

a first input configured to receive an estimated glucose level in real time;

a second input configured to receive a reference glucose level;

a Kalman control algorithm configured to determine a control command based on the estimated glucose level and the reference glucose level, wherein the Kalman control algorithm has a dynamic process model forced by the control command and a cost function defining a desired level of control; and

an output configured to provide the control command to a pump, wherein the pump provides medication in response to the control command to minimize a difference between the estimated glucose level and the reference glucose level.

53. An artificial pancreas for controlling glucose level in real time comprising:

a glucose sensor to provide a measurement of the glucose level;

an optimal glucose estimator, wherein the optimal glucose estimator uses a stochastic model to describe a physiological process relating to the glucose level and uses a linearized Kalman filter to estimate the glucose level in real time based on the measurement from the glucose sensor;

an optimal glucose controller, wherein the optimal glucose controller uses a substantially identical stochastic model as the optimal glucose estimator and uses a Kalman control algorithm to determine a control command to adjust the glucose level in real time; and

a medical dispenser to provide medication to a patient in response to the control command.

54. The artificial pancreas of claim 53, wherein the optimal glucose estimator and the optimal glucose controller are implemented using a software algorithm.

55. The artificial pancreas of claim 53, wherein the artificial pancreas is a portable device.

56. The artificial pancreas of claim 53 further comprising one or more additional sensors of different types which operate independently to provide respective glucose measurements.

57. The artificial pancreas of claim 53, wherein the medical dispenser secretes insulin and glucagon to control a relatively high glucose level and a relatively low glucose level respectively.

58. The artificial pancreas of claim 53 further comprising a patient health monitor with an input/output interface to receive inputs from a user and to display status of the artificial pancreas.