Adaptive signal processing

Abstract

Disclosed herein are techniques related to adaptive signal processing. The techniques may involve: obtaining a plurality of unfiltered measurement values based on signals generated by a sensor; determining a plurality of filtered measurement values based on the plurality of unfiltered measurement values; determining, based on the plurality of filtered measurements, a first derivative metric for a current filtered measurement of the plurality of filtered measurements and a second derivative metric for the current filtered measurement; determining an output filtered measurement indicative of a physiological condition of a user based at least in part on the current filtered measurement, the first derivative metric, the second derivative metric, and a previous output measurement; and outputting the output filtered measurement.

What is claimed is: 
     
         1 . A system for controlling insulin delivery, the system comprising:
 one or more processors; and   one or more processor-readable storage media storing instructions which, when executed by the one or more processors, cause performance of:
 obtaining a plurality of unfiltered measurement values based on signals generated by a sensor; 
 determining a plurality of filtered measurement values based on the plurality of unfiltered measurement values; 
 determining, based on the plurality of filtered measurements, a first derivative metric for a current filtered measurement of the plurality of filtered measurements and a second derivative metric for the current filtered measurement; 
 determining an output filtered measurement indicative of a physiological condition of a user based at least in part on the current filtered measurement, the first derivative metric, the second derivative metric, and a previous output measurement; and 
 outputting the output filtered measurement. 
   
     
     
         2 . The system of  claim 1 , wherein outputting the output filtered measurement comprises outputting the output filtered measurement for transmission to an infusion device for delivery of insulin based on the output filtered measurement. 
     
     
         3 . The system of  claim 1 , wherein the one or more processor-readable storage media further store instructions which, when executed by the one or more processors, cause performance of:
 determining a process variance metric based at least in part on the first derivative metric and the second derivative metric,   wherein determining the output filtered measurement comprises determining the output filtered measurement indicative of the physiological condition of the user based at least in part on the current filtered measurement, the process variance, and the previous output measurement.   
     
     
         4 . The system of  claim 1 , wherein the one or more processor-readable storage media further store instructions which, when executed by the one or more processors, cause performance of:
 determining a frequency estimate associated with the current filtered measurement based on the first derivative metric; and   determining a noise estimate associated with the current filtered measurement based on the second derivative metric,   wherein determining the output filtered measurement comprises determining the output filtered measurement indicative of the physiological condition of the user based at least in part on the current filtered measurement, the frequency estimate, the noise estimate, and the previous output measurement.   
     
     
         5 . The system of  claim 4 ,
 wherein determining the frequency estimate comprises scaling the first derivative metric by a calibration factor for converting the output filtered measurement to a second value, the first derivative metric comprising an average of first derivative values associated with the current filtered measurement and one or more preceding filtered measurements,   wherein determining the noise estimate comprises scaling the second derivative metric by the calibration factor, the second derivative metric comprising an average of second derivative values associated with eh current filtered measurement and the one or more preceding filtered measurements.   
     
     
         6 . The system of  claim 4 , wherein the one or more processor-readable storage media further store instructions which, when executed by the one or more processors, cause performance of:
 determining a rate of change metric associated with the current filtered measurement based at least in part on the first derivative metric;   scaling the rate of change metric based on the noise estimate to generate a scaled rate of change metric; and   adding the scaled rate of change metric to the current filtered measurement to generate an adjusted filtered measurement,   wherein determining the output filtered measurement comprises determining the output filtered measurement based at least in part on the adjusted filtered measurement, the frequency estimate, the noise estimate, and the previous output measurement.   
     
     
         7 . The system of  claim 1 , wherein the one or more processor-readable storage media further store instructions which, when executed by the one or more processors, cause performance of:
 identifying a dropout condition based at least in part on one or more of the first derivative metric and the second derivative metric associated with the current filtered measurement; and   determining an adjusted filtered measurement in response to identifying the dropout condition,   wherein determining the output filtered measurement comprises determining the output filtered measurement based at least in part on the adjusted filtered measurement, the first derivative metric, the second derivative metric, and the previous output measurement.   
     
     
         8 . The system of  claim 7 , wherein identifying the dropout condition comprises identifying the dropout condition when the first derivative metric associated with the current filtered measurement is greater than a first derivative dropout threshold value and the second derivative metric associated with the current filtered measurement is less than a second derivative dropout threshold value. 
     
     
         9 . The system of  claim 1 , wherein the one or more processor-readable storage media further store instructions which, when executed by the one or more processors, cause performance of:
 adjusting the current filtered measurement to compensate for delay based at least in part on one or more of the first derivative metric and the second derivative metric associated with the current filtered measurement, resulting in an adjusted filtered measurement,   wherein determining the output filtered measurement comprises determining the output filtered measurement based on the adjusted filtered measurement, the first derivative metric, the second derivative metric, and the previous output measurement.   
     
     
         10 . The system of  claim 9 , wherein the one or more processor-readable storage media further store instructions which, when executed by the one or more processors, cause performance of:
 determining a noise estimate associated with the current filtered measurement based on the second derivative metric associated with the current filtered measurement,   wherein adjusting the current filtered measurement comprises:
 scaling a rate of change metric for the current filtered measurement based on the noise estimate, resulting in a scaled rate of change metric; and 
 adding the scaled rate of change metric to the current filtered measurement to obtain the adjusted filtered measurement. 
   
     
     
         11 . The system of  claim 1 , wherein determining the output filtered measurement comprises:
 determining an intermediate error estimate based on a preceding output error estimate, the first derivative metric, and the second derivative metric;   determining a filter gain value based on the intermediate error estimate and a measurement error value; and   determining the output filtered measurement based on the current filtered measurement, the filter gain value, and the previous output measurement.   
     
     
         12 . The system of  claim 1 , wherein determining the output filtered measurement comprises:
 implementing a Kalman filter to filter the current filtered measurement using the previous output measurement, the first derivative metric, and the second derivative metric.   
     
     
         13 . The system of  claim 12 , wherein implementing the Kalman filter comprises:
 determining an intermediate error estimate based on a preceding output error estimate, the first derivative metric, and the second derivative metric;   determining a Kalman filter gain value based on the intermediate error estimate and a measurement error value; and   determining the output filtered measurement according to the equation i out =i out [n−1]+k(i sig −i out [n−1]), where i out  is the output measurement value, i out [n−1] represents the preceding output measurement, i sig  represents the current filtered measurement, and k represents the Kalman filter gain value.   
     
     
         14 . A processor-implemented method for controlling insulin delivery, the method comprising:
 obtaining a plurality of unfiltered measurement values based on signals generated by a sensor;   determining a plurality of filtered measurement values based on the plurality of unfiltered measurement values;   determining, based on the plurality of filtered measurements, a first derivative metric for a current filtered measurement of the plurality of filtered measurements and a second derivative metric for the current filtered measurement;   determining an output filtered measurement indicative of a physiological condition of a user based at least in part on the current filtered measurement, the first derivative metric, the second derivative metric, and a previous output measurement; and   outputting the output filtered measurement.   
     
     
         15 . The method of  claim 14 , wherein outputting the output filtered measurement comprises outputting the output filtered measurement for transmission to an infusion device for delivery of insulin based on the output filtered measurement. 
     
     
         16 . The method of  claim 14 , further comprising:
 determining a process variance metric based at least in part on the first derivative metric and the second derivative metric,   wherein determining the output filtered measurement comprises determining the output filtered measurement indicative of the physiological condition of the user based at least in part on the current filtered measurement, the process variance, and the previous output measurement.   
     
     
         17 . The method of  claim 14 , further comprising:
 determining a frequency estimate associated with the current filtered measurement based on the first derivative metric; and   determining a noise estimate associated with the current filtered measurement based on the second derivative metric,   wherein determining the output filtered measurement comprises determining the output filtered measurement indicative of the physiological condition of the user based at least in part on the current filtered measurement, the frequency estimate, the noise estimate, and the previous output measurement.   
     
     
         18 . The method of  claim 14 , further comprising:
 identifying a dropout condition based at least in part on one or more of the first derivative metric and the second derivative metric associated with the current filtered measurement; and   determining an adjusted filtered measurement in response to identifying the dropout condition,   wherein determining the output filtered measurement comprises determining the output filtered measurement based at least in part on the adjusted filtered measurement, the first derivative metric, the second derivative metric, and the previous output measurement.   
     
     
         19 . The method of  claim 14 , further comprising:
 adjusting the current filtered measurement to compensate for delay based at least in part on one or more of the first derivative metric and the second derivative metric associated with the current filtered measurement, resulting in an adjusted filtered measurement,   wherein determining the output filtered measurement comprises determining the output filtered measurement based on the adjusted filtered measurement, the first derivative metric, the second derivative metric, and the previous output measurement.   
     
     
         20 . One or more non-transitory processor-readable storage media storing instructions which, when executed by one or more processors, cause performance of:
 obtaining a plurality of unfiltered measurement values based on signals generated by a sensor;   determining a plurality of filtered measurement values based on the plurality of unfiltered measurement values;   determining, based on the plurality of filtered measurements, a first derivative metric for a current filtered measurement of the plurality of filtered measurements and a second derivative metric for the current filtered measurement;   determining an output filtered measurement indicative of a physiological condition of a user based at least in part on the current filtered measurement, the first derivative metric, the second derivative metric, and a previous output measurement; and   outputting the output filtered measurement.