Method and apparatus for real-time estimation of physiological parameters
First Claim
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 Assignments
0 Petitions
Accused Products
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.
554 Citations
34 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. - View Dependent Claims (2, 3, 4, 5, 6)
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 real-time 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. - View Dependent Claims (8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24)
-
-
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. - View Dependent Claims (26, 27, 28, 29, 30)
-
-
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 real-time estimate of the physiological parameter. - View Dependent Claims (32, 33, 34)
-
Specification