Sensor error mitigation

Abstract

Techniques disclosed herein relate generally to sensor error mitigation. In some embodiments, the techniques involve identifying an error metric associated with an input variable to a translation model, determining a reference output of the translation model by providing a reference input value for the input variable to the translation model, generating a modulated value for the input variable based on the reference input value using the error metric, determining a simulated output of the translation model by providing the modulated value for the input variable to the translation model, and updating the translation with a reduced weighting applied to the input variable when a difference between the simulated output and the reference output is greater than a threshold.

What is claimed is: 
     
         1 . A system comprising:
 one or more processors; and   one or more processor-readable media storing instructions which, when executed by the one or more processors, cause performance of:
 identifying an error metric associated with an input variable to a translation model, the translation model providing outputs influenced by values for the input variable and a weighting applied to the input variable; 
 determining a reference output of the translation model by providing a reference input value for the input variable to the translation model; 
 generating a modulated value for the input variable based on the reference input value using the error metric; 
 determining a simulated output of the translation model by providing the modulated value for the input variable to the translation model; and 
 updating the translation model with a reduced weighting applied to the input variable when a difference between the simulated output and the reference output is greater than a threshold, resulting in an updated translation model which mitigates error associated with the input variable. 
   
     
     
         2 . The system of  claim 1 , wherein:
 the translation model comprises an estimation model for providing an estimated glucose value;   determining the reference output comprises determining a reference glucose value using the reference input value;   determining the simulated output comprises determining a simulated glucose value using the modulated value; and   updating the translation model comprises updating the estimation model to reduce the difference between the simulated glucose value and the reference glucose value by assigning the reduced weighting to the input variable in the estimation model.   
     
     
         3 . The system of  claim 2 , wherein:
 the estimation model comprises a sensor glucose estimation model for providing an estimated glucose value;   the reference glucose value comprises an estimated glucose value determined by the sensor glucose estimation model as a function of the reference input value and the weighting applied to the input variable; and   the simulated glucose value comprises an estimated glucose value determined by the sensor glucose estimation model as a function of the modulated value and the weighting applied to the input variable.   
     
     
         4 . The system of  claim 2 , wherein the translation model comprises an insight model for generating a notification. 
     
     
         5 . The system of  claim 4 , wherein updating the translation model comprises:
 updating the insight model for generating the notification as a function of the estimated glucose value provided by the updated estimation model having the reduced weighting applied to the input variable.   
     
     
         6 . The system of  claim 1 , wherein updating the translation model comprises:
 reducing the weighting applied to the input variable;   determining an updated reference output based at least in part on the reduced weighting; and   determining an updated simulated output based at least in part on the reduced weighting applied to the modulated value such that the difference between the simulated output and the reference output is less than the threshold.   
     
     
         7 . The system of  claim 1 , wherein the input variable comprises an output electrical current, an electrode voltage, or an electrochemical impedance spectroscopy (EIS) value determined based on electrical signals provided by an interstitial glucose sensing arrangement. 
     
     
         8 . A processor-implemented method comprising:
 identifying an error metric associated with an input variable to a translation model, the translation model providing outputs influenced by values for the input variable and a weighting applied to the input variable;   determining a reference output of the translation model by providing a reference input value for the input variable to the translation model;   generating a modulated value for the input variable based on the reference input value using the error metric;   determining a simulated output of the translation model by providing the modulated value for the input variable to the translation model; and   updating the translation model with a reduced weighting applied to the input variable when a difference between the simulated output and the reference output is greater than a threshold, resulting in an updated translation model which mitigates error associated with the input variable.   
     
     
         9 . The processor-implemented method of  claim 8 , wherein:
 the translation model comprises an estimation model for providing an estimated glucose value;   determining the reference output comprises determining a reference glucose value using the reference input value;   determining the simulated output comprises determining a simulated glucose value using the modulated value; and   updating the translation model comprises updating the estimation model to reduce the difference between the simulated glucose value and the reference glucose value by assigning the reduced weighting to the input variable in the estimation model.   
     
     
         10 . The processor-implemented method of  claim 9 , wherein:
 the estimation model comprises a sensor glucose estimation model for providing an estimated glucose value;   the reference glucose value comprises an estimated glucose value determined by the sensor glucose estimation model as a function of the reference input value and the weighting applied to the input variable; and   the simulated glucose value comprises an estimated glucose value determined by the sensor glucose estimation model as a function of the modulated value and the weighting applied to the input variable.   
     
     
         11 . The processor-implemented method of  claim 9 , wherein the translation model comprises an insight model for generating a notification. 
     
     
         12 . The processor-implemented method of  claim 11 , wherein updating the translation model comprises:
 updating the insight model for generating the notification as a function of the estimated glucose value provided by the updated estimation model having the reduced weighting applied to the input variable.   
     
     
         13 . The processor-implemented method of  claim 8 , wherein updating the translation model comprises:
 reducing the weighting applied to the input variable;   determining an updated reference output based at least in part on the reduced weighting; and   determining an updated simulated output based at least in part on the reduced weighting applied to the modulated value such that the difference between the simulated output and the reference output is less than the threshold.   
     
     
         14 . The processor-implemented method of  claim 8 , wherein the input variable comprises an output electrical current, an electrode voltage, or an electrochemical impedance spectroscopy (EIS) value determined based on electrical signals provided by an interstitial glucose sensing arrangement. 
     
     
         15 . One or more non-transitory processor-readable media storing instructions which, when executed by one or more processors, cause performance of:
 identifying an error metric associated with an input variable to a translation model, the translation model providing outputs influenced by values for the input variable and a weighting applied to the input variable;   determining a reference output of the translation model by providing a reference value for the input variable to the translation model;   generating a modulated value for the input variable based on the reference value using the error metric;   determining a simulated output of the translation model by providing the modulated value for the input variable to the translation model; and   updating the translation model with a reduced weighting applied to the input variable when a difference between the simulated output and the reference output is greater than a threshold, resulting in an updated translation model which mitigates error associated with the input variable.   
     
     
         16 . The one or more non-transitory processor-readable media of  claim 15 , wherein:
 the translation model comprises an estimation model for providing an estimated glucose value;   determining the reference output comprises determining a reference glucose value using the reference input value;   determining the simulated output comprises determining a simulated glucose value using the modulated value; and   updating the translation model comprises updating the estimation model to reduce the difference between the simulated glucose value and the reference glucose value by assigning the reduced weighting to the input variable in the estimation model.   
     
     
         17 . The one or more non-transitory processor-readable media of  claim 16 , wherein:
 the estimation model comprises a sensor glucose estimation model for providing an estimated glucose value;   the reference glucose value comprises an estimated glucose value determined by the sensor glucose estimation model as a function of the reference input value and the weighting applied to the input variable; and   the simulated glucose value comprises an estimated glucose value determined by the sensor glucose estimation model as a function of the modulated value and the weighting applied to the input variable.   
     
     
         18 . The one or more non-transitory processor-readable media of  claim 16 , wherein the translation model comprises an insight model for generating a notification. 
     
     
         19 . The one or more non-transitory processor-readable media of  claim 18 , wherein updating the translation model comprises:
 updating the insight model for generating the notification as a function of the estimated glucose value provided by the updated estimation model having the reduced weighting applied to the input variable.   
     
     
         20 . The one or more non-transitory processor-readable media of  claim 15 , wherein updating the translation model comprises:
 reducing the weighting applied to the input variable;   determining an updated reference output based at least in part on the reduced weighting; and   determining an updated simulated output based at least in part on the reduced weighting applied to the modulated value such that the difference between the simulated output and the reference output is less than the threshold.