Real time self-adjusting calibration algorithm
A method of calibrating glucose monitor data includes collecting the glucose monitor data over a period of time at predetermined intervals. It also includes obtaining at least two reference glucose values from a reference source that temporally correspond with the glucose monitor data obtained at the predetermined intervals. Also included is calculating the calibration characteristics using the reference glucose values and corresponding glucose monitor data to regress the obtained glucose monitor data. And, calibrating the obtained glucose monitor data using the calibration characteristics is included. In preferred embodiments, the reference source is a blood glucose meter, and the at least two reference glucose values are obtained from blood tests. In additional embodiments, calculation of the calibration characteristics includes linear regression and, in particular embodiments, least squares linear regression. Alternatively, calculation of the calibration characteristics includes non-linear regression. Data integrity may be verified and the data may be filtered.
1. A method for filtering signals from a sensor comprising:
receiving a plurality of signal values from the sensor generated in response to a concentration of a substance;
at a processor, computing a quantifier of a variance of a measurement error based, at least in part, on a standard deviation of an absolute value of consecutive data points within the plurality of signal values; and
at said processor, filtering the plurality of signal values with an adaptive filter, wherein the quantifier is an input to the adaptive filter.
2. The method of claim 1 , where the sensor comprises a glucose sensor.
3. The method of claim 2 , wherein the plurality of data values are responsive to blood glucose concentrations.
4. The method of claim 1 , wherein formulating the standard deviation comprises formulating a windowed, unweighted standard deviation.
5. The method of claim 4 , wherein the quantifier is equal to
R k =c*δ k +b,
where
δ
k
=
∑
i
=
k
-
l
k
(
s
i
-
s
i
-
l
-
∑
i
=
k
-
l
k
s
i
-
s
i
-
1
l
+
1
)
2
l
,
s k is a raw signal sampled at a k th discrete time interval and l is a window size of a history consecutive differences of the raw signal, and wherein c and b are constants.
6. The method of claim 1 , wherein the standard deviation is formulated as a recursive, weighted standard deviation.
7. The method of claim 6 , wherein the quantifier is equal to
R k =c*δ k +b,
where
δ
k
=
∑
(
α
k
*
(
s
k
-
s
k
-
1
-
∑
(
α
k
*
s
k
-
s
k
-
1
)
∑
α
k
)
2
)
∑
α
k
-
1
,
s k is a raw signal sampled at a k th discrete time interval, α k is a growing exponential weight: α k =e ((k*Δt/60)/τ) , τ is an exponential time constant in hours and Δt is a sampling time in minutes, and wherein c and b are constants.
8. The method of claim 1 , wherein the adaptive filter comprises a Kalman filter.
9. A method for calibrating a sensor comprising:
receiving a plurality of signal values from the sensor generated in response to a concentration of a substance;
computing, at a processor, a reliability metric for each signal value of the plurality of data values based, at least in part, on a standard deviation of an absolute value of consecutive data points within the plurality of signal values;
selectively discarding each of one or more of the signal values of the plurality of signal values in response to application of a threshold value to said reliability metric for the signal value;
filtering, at said processor, the signal values of the plurality of signal values that have not been discarded using an adaptive filter; and
adjusting an output of the sensor using the filtered signal values.
10. A method for calibrating a sensor comprising:
receiving a plurality of signal values from the sensor responsive to a concentration of a substance;
computing, at a processor, a reliability of each signal value of the plurality of signal values;
selectively discarding each of one Or more of the signal values of the plurality of signal values in response to application of a threshold value to said computed reliability metric for the signal;
filtering, at said processor, the signal values of the plurality of signal values that have not been discarded; and
adjusting an output of the sensor using the filtered signal values;
wherein filtering the signal values comprises filtering the signal values with an adaptive filter;
wherein filtering the signal values with an adaptive filter comprises using the adaptive filter with a parameter based, at least in part, on the signal values of the plurality of signal values that have not been discarded; and
wherein the parameter is representative of a standard deviation of the signal values of the plurality of signal values that have not been discarded or is representative of a standard deviation of an absolute value of signal values within the signal values of the plurality of signal values that have not been discarded.
11. The method of claim 10 , wherein the standard deviation comprises a windowed, unweighted standard deviation.
12. The method of claim 10 , wherein the standard deviation comprises a recursive, weighted standard deviation.
13. An apparatus for filtering signal values from a sensor comprising:
means, comprising a processor, for receiving a plurality of signal values from the sensor;
means, comprising said processor, for Obtaining a quantifier of a variance of a measurement error associated with the plurality of signal values based, at least in part, on a standard deviation of an absolute value of consecutive data points within the plurality of signal values; and
means, comprising said processor, for filtering the plurality of signal values with an adaptive filter, wherein the quantifier is an input value to the adaptive filter.
14. The apparatus of claim 13 , where the sensor comprises a glucose sensor.
15. The apparatus of claim 14 , wherein the plurality of signal ues are generated by said sensor in response to blood glucose concentrations.
16. The apparatus of claim 13 , wherein said means for obtaining a quantifier comprises means for formulating a standard deviation of an absolute value of consecutive signal values within the plurality of signal values.
17. The apparatus of claim 16 , wherein said means for formulating the standard deviation comprises means for formulating a windowed, unweighted standard deviation.
18. The apparatus of claim 16 , wherein said means for formulating the standard deviation comprises means for formulating a recursive, weighted standard deviation.
19. The apparatus of claim 13 , wherein the adaptive filter comprises a Kalman filter.
20. The method of claim 9 , wherein the adaptive filter comprises a Kalman filter.