Sensor signal processing with kalman-based calibration

Abstract

Method, system, and computer readable media are disclosed for estimating an amount of an analyte in blood of a subject. In an example, a method includes obtaining a noise filtered current by noise filtering a current received from an analyte sensor placed interstitially in tissue of the subject, estimating one or more blood analyte calibration parameters based on the noise filtered current, obtaining a second set of one or more interstitial analyte calibration parameters based at least in part on the one or more blood analyte calibration parameters, and estimating the amount of the analyte in the blood of the subject based on the noise filtered current and the one or more interstitial analyte calibration parameters. In this way, an analyte sensor that is placed interstitially may readily be calibrated to report accurate blood analyte values.

What is claimed is: 
     
         1 . A method of estimating an amount of an analyte in blood of a subject, comprising:
 receiving, from an analyte sensor placed an interstitial fluid of the subject, a current reflective of an interstitial level of the analyte;   noise filtering the current to provide a noise filtered current;   estimating a first set of one or more blood analyte calibration parameters based at least in part on the noise filtered current;   obtaining a second set of one or more interstitial analyte calibration parameters based at least in part on the first set of one or more blood analyte calibration parameters; and   estimating the amount of the analyte in the blood of the subject based on the second set of one or more interstitial analyte calibration parameters and the noise filtered current.   
     
     
         2 . The method of  claim 1 , wherein noise filtering the current values further comprises filtering out both a low frequency high magnitude noise component and a high frequency low magnitude component, where the low frequency high magnitude noise component is filtered heuristically and where the high frequency low magnitude component is filtered via Kalman filtering. 
     
     
         3 . The method of  claim 1 , wherein estimating the first set of one or more blood analyte calibration parameters further comprises:
 deconvoluting the noise filtered current via a sequential Kalman filter to obtain a hypothetical analyte sensor current assuming the analyte sensor is inserted in the blood instead of the interstitial fluid; and   estimating the first set of one or more blood glucose calibration parameters based on the hypothetical analyte sensor current using a Kalman filter that uses the hypothetical analyte sensor current as a measurement variable.   
     
     
         4 . The method of  claim 3 , further comprising obtaining a capillary blood analyte measurement; and
 wherein estimating the first set of one or more blood analyte calibration parameters is based on the capillary blood glucose measurement.   
     
     
         5 . The method of  claim 3 , further comprising determining an initial blood analyte calibration parameter state estimate, and a blood analyte calibration parameter covariance, for initialization of the sequential Kalman filter. 
     
     
         6 . The method of  claim 5 , wherein the initial blood analyte calibration parameter state estimate and the blood analyte calibration parameter covariance is determined using a nonlinear Kalman filter. 
     
     
         7 . The method of  claim 6 , wherein the nonlinear Kalman filter is a cubature Kalman filter. 
     
     
         8 . The method of  claim 5 , further comprising determining the initial blood analyte calibration parameter state estimate and the blood analyte calibration parameter covariance based on one or more training data sets. 
     
     
         9 . The method of  claim 5 , further comprising determining the initial blood analyte calibration parameter state estimate and the blood analyte calibration parameter covariance arbitrarily. 
     
     
         10 . The method of  claim 5 , further comprising determining the initial blood analyte calibration parameter state estimate and the blood analyte calibration parameter covariance based on one or more prior analyte sensor calibration events. 
     
     
         11 . The method of  claim 1 , wherein the analyte sensor is a glucose sensor and the analyte is glucose. 
     
     
         12 . The method of  claim 11 , further comprising:
 sending one or more instructions to an insulin delivery system for controlling operation of an insulin pump that in turn delivers an appropriate amount of insulin to the subject, the one or more instructions based on the estimated amount of analyte in the blood of the subject.   
     
     
         13 . The method of  claim 11 , further comprising displaying the estimated amount of glucose in the blood of the subject on a display screen associated with the glucose sensor, for viewing by the subject. 
     
     
         14 . A system for estimating an amount of an analyte in a subject, comprising:
 an analyte sensor; and   a mobile computing device, the mobile computing device including a processor storing instructions in non-transitory memory that, when executed, cause the processor to:   receive from the analyte sensor a current reflective of an interstitial level of the analyte in the subject;   noise filter the current to obtain a noise filtered current;   estimate one or more blood analyte calibration parameters based on the noise filtered current;   obtain one or more interstitial analyte calibration parameters based on the one or more blood analyte calibration parameters; and
 estimate the amount of the analyte in the blood of the subject based on the one or more interstitial analyte calibration parameters and the noise filtered current. 
   
     
     
         15 . The system of  claim 14 , wherein the processor stores further instructions in non-transitory memory that, when executed, cause the processor to:
 obtain the noise filtered current by heuristically filtering out a low frequency high magnitude noise component of the current, and by Kalman-based filtering of a high frequency low magnitude noise component of the current.   
     
     
         16 . The system of  claim 14 , wherein the processor stores further instructions in non-transitory memory that, when executed, cause the processor to:
 deconvolute the noise filtered current via a sequential Kalman filter to obtain a predicted blood analyte sensor current; and   estimate the one or more blood glucose calibration parameters based on the predicted blood analyte sensor current via a Kalman filter that uses the predicted analyte sensor current as a measurement variable.   
     
     
         17 . The system of  claim 16 , wherein the processor stores further instructions in non-transitory memory that, when executed, cause the processor to:
 estimate the one or more blood glucose calibration parameters based on a capillary blood glucose measurement.   
     
     
         18 . The system of  claim 16 , wherein the processor stores further instructions in non-transitory memory that, when executed, cause the processor to:
 determine an initial blood analyte calibration parameter state estimate, and an initial blood analyte calibration parameter covariance; and
 initialize the sequential Kalman filter based on the initial blood analyte calibration parameter state estimate and the initial blood analyte calibration parameter covariance. 
   
     
     
         19 . The system of  claim 18 , wherein the processor stores further instructions in non-transitory memory that, when executed, cause the processor to:
 determine the initial blood analyte calibration parameter state estimate, and the initial blood analyte calibration parameter covariance via a nonlinear Kalman filter.   
     
     
         20 . The system of  claim 19 , wherein the nonlinear Kalman filter is a cubature Kalman filter. 
     
     
         21 . The system of  claim 18 , wherein the processor stores further instructions in non-transitory memory that, when executed, cause the processor to:
 arbitrarily assign the initial blood analyte calibration parameter state estimate, and the initial blood analyte calibration parameter covariance.   
     
     
         22 . The system of  claim 18 , wherein the processor stores further instructions in non-transitory memory that, when executed, cause the processor to:
 determine the initial blood analyte calibration parameter state estimate and the initial blood analyte calibration parameter covariance based on one or more prior analyte sensor calibration events.   
     
     
         23 . The system of  claim 18 , wherein the processor stores further instructions in non transitory memory that, when executed, cause the processor to:
 determine the initial blood analyte calibration parameter state estimate and the initial blood analyte calibration parameter covariance based on one or more training data sets.   
     
     
         24 . The system of  claim 14 , wherein the analyte sensor is a glucose sensor and wherein the analyte is glucose. 
     
     
         25 . The system of  claim 24 , further comprising:
 an insulin delivery unit that includes at least an insulin pump and an infusion set; and   wherein the processor stores further instructions in non-transitory memory that, when executed, cause the processor to:
 determine an insulin injection amount for the subject based on the estimated amount of the analyte in the blood of the subject; and 
 send one or more instructions to the insulin delivery unit to control the insulin pump to delivery the insulin injection amount to the subject via the infusion set. 
   
     
     
         26 . The system of  claim 24 , wherein the mobile computing device further comprises:
 a display; and   wherein the processor stores further instructions in non-transitory memory that, when executed, cause the processor to:
 provide the amount of the analyte in the blood of the subject in a viewable form on the display, for viewing by the subject. 
   
     
     
         27 . A non-transitory computer-readable storage medium with an executable program stored thereon for calibrating an analyte sensor inserted into the tissue of a subject, wherein the program instructs a microprocessor to perform the steps of:
 receiving from the analyte sensor a current reflective of an interstitial amount of an analyte in the subject;   noise filtering the current received from the analyte sensor to provide a noise filtered current;   estimating a first set of one or more blood analyte calibration parameters based on the noise filtered current;   obtaining a second set of one or more interstitial analyte calibration parameters based on the first set of one or more blood analyte calibration parameters; and
 estimating an amount of the analyte in the blood of the subject based on the second set of one or more interstitial analyte calibration parameters and the noise filtered current. 
   
     
     
         28 . The non-transitory computer-readable storage medium of  claim 27 , wherein the program further instructs the microprocessor to filter out low frequency high magnitude noise heuristically and filter out high frequency low magnitude noise via Kalman filtering. 
     
     
         29 . The non-transitory computer-readable storage medium of  claim 27 , wherein the program further instructs the microprocessor to estimate the first set of one or more blood analyte calibration parameters by deconvoluting the noise filtered current via a sequential Kalman filter to obtain a predicted blood analyte sensor current that accounts for interstitial-to-blood analyte diffusion dynamics; and
 estimate the first set of one or more blood analyte calibration parameters based on the predicted blood analyte sensor current via a Kalman filter that uses the hypothetical analyte sensor current as a measurement variable.   
     
     
         30 . The non-transitory computer-readable storage medium of  claim 29 , wherein the program further instructs the microprocessor to estimate the first set of one or more blood analyte calibration parameters based at least in part on one or more capillary blood analyte measurements. 
     
     
         31 . The non-transitory computer-readable storage medium of  claim 29 , wherein the program further instructs the microprocessor to determine an initial blood analyte calibration parameter state estimate, and an initial blood analyte calibration parameter covariance, for initiation of the sequential Kalman filter. 
     
     
         32 . The non-transitory computer-readable storage medium of  claim 31 , wherein the program further instructs the microprocessor to determine the initial blood analyte calibration parameter state estimate, and the initial blood analyte calibration parameter covariance via a nonlinear Kalman filter. 
     
     
         33 . The non-transitory computer-readable storage medium of  claim 32 , wherein the nonlinear Kalman filter is a cubature Kalman filter. 
     
     
         34 . The non-transitory computer-readable storage medium of  claim 31 , wherein the program further instructs the microprocessor to determining the initial blood analyte calibration parameter state estimate and the blood analyte parameter covariance based on one or more training data sets. 
     
     
         35 . The non-transitory computer-readable storage medium of  claim 31 , wherein the program further instructs the microprocessor to arbitrarily determine the initial blood analyte calibration parameter state estimate and the blood analyte parameter covariance. 
     
     
         36 . The non-transitory computer-readable storage medium of  claim 31 , wherein the program further instructs the microprocessor to determine the initial blood analyte calibration parameter state estimate and the blood analyte parameter covariance based on one or more prior analyte sensor calibration events. 
     
     
         37 . The non-transitory computer-readable storage medium of  claim 27 , wherein the analyte sensor is a glucose sensor; and
 wherein the analyte is glucose.   
     
     
         38 . The non-transitory computer-readable storage medium of  claim 37 , wherein the program further instructs the microprocessor to determine an insulin injection amount for the subject based on the estimate of the estimated amount of the analyte in the blood of the subject; and
 provide one or more instructions to pertaining to controlling an insulin pump based on the insulin injection amount, the instructions for use via the insulin pump.   
     
     
         39 . The non-transitory computer-readable storage medium of  claim 37 , wherein the program further instructs the microprocessor to provide one or more instructions for displaying the estimated amount of the analyte in the blood of the subject on a display viewable by the subject.