End of life detection for analyte sensors
Systems and methods for processing sensor data and end of life detection are provided. In some embodiments, a method for determining the end of life of a continuous analyte sensor includes evaluating a plurality of risk factors using an end of life function to determine an end of life status of the sensor and providing an output related to the end of life status of the sensor. The plurality of risk factors may be selected from the list including the number of days the sensor has been in use, whether there has been a decrease in signal sensitivity, whether there is a predetermined noise pattern, whether there is a predetermined oxygen concentration pattern, and error between reference BG values and EGV sensor values.
1 . A method for detecting outliers in data, comprising:
generating sensor data comprising a signal from an analyte sensor at least partially implanted under a skin of a patient;
storing at least one calibration data set in a memory;
evaluating the at least one calibration data set comprising external reference values;
determining a boundary associated with the at least one calibration data set, wherein the boundary associated with the at least one calibration data set is determined after the sensor is at least partially implanted under the skin of the patient;
determining at least one value outside the boundary as a possible outlier;
evaluating a relevancy of the possible outlier to determine outlier information, wherein the evaluating includes receiving input from the patient indicative of clinical context relating to the possible outlier, the patient input comprising at least one of: meal intake, insulin administration, or exercise to determine a cause of the possible outlier;
performing processing responsive to the outlier information, wherein the processing includes transforming the at least one calibration data set by removing the possible outlier based on the outlier information;
calculating a value representative of an analyte concentration based on the at least one calibration data set after the performing the processing responsive to the outlier information; and
transmitting the value representative of the analyte concentration to a display.
2 . The method of claim 1 , wherein the boundary is a confidence interval.
3 . The method of claim 1 , wherein the evaluating the at least one calibration data set comprises determining a best subset.
4 . The method of claim 3 , wherein the determining the best subset includes generating a regression line.
5 . The method of claim 4 , wherein the regression line is generated using at least half of data points in the at least one calibration data set.
6 . The method of claim 3 , wherein the determining the best subset includes generating a convex hull.
7 . The method of claim 1 , wherein the evaluating the relevancy of the possible outlier to determine outlier information comprises examining one or more factors from the following list: (1) an amplitude of error of a data point relative to a best line, (2) the direction of error a data point relative to the best line, (3) a clinical risk of the data at a time stamp of the data point, (4) a rate of change of an analyte concentration or derivative of the sensor data associated with a data point, (5) a rate of acceleration or deceleration of the analyte concentration or second derivative of the sensor data associated with the data point.
8 . The method of claim 1 , wherein the performing processing responsive to the outlier information comprises removing the possible outlier from the calibration data set.
9 . The method of claim 8 , wherein the removing the possible outlier is performed prospectively.
10 . The method of claim 8 , wherein the removing the possible outlier is performed retrospectively.
11 . The method of claim 8 , wherein the removing the possible outlier comprises removing the outlier temporarily from the calibration data set.
12 . The method of claim 8 , wherein the removing the possible outlier comprises removing the outlier permanently from the calibration data set.
13 . A method for detecting outliers in data, comprising:
generating sensor data comprising a signal from an analyte sensor at least partially implanted under a skin of a patient;
evaluating a plurality of subsets of a calibration data set to determine a best subset;
identifying a boundary associated with the best subset, wherein the boundary associated with the best subset is determined in real-time after the sensor is at least partially implanted under the skin of the patient;
identifying at least one value outside the boundary as a possible outlier;
evaluating a relevancy of the possible outlier to determine outlier information, wherein the evaluating includes receiving input from the patient indicative of a cause relating to the possible outlier, the patient input comprising at least one of: meal intake, insulin administration, or physical activity;
performing processing responsive to the outlier information, wherein the processing includes modifying the best subset by removing the possible outlier based on the outlier information;
calculating a value representative of an analyte concentration based on the best subset after the performing the processing responsive to the outlier information; and
transmitting the value representative of the analyte concentration to a display.
14 . The method of claim 13 , wherein the determining the best subset includes generating a regression line.
15 . The method of claim 13 , wherein the boundary is associated with a confidence interval.
16 . The method of claim 13 , wherein the evaluating the relevancy of the possible outlier to determine outlier information comprises examining one or more factors from the following list: (1) an amplitude of error of a data point relative to a best line, (2) the direction of error a data point relative to the best line, (3) a clinical risk of the data at a time stamp of the data point, (4) a rate of change of an analyte concentration or derivative of the sensor data associated with a data point, (5) a rate of acceleration or deceleration of the analyte concentration or second derivative of the sensor data associated with the data point.
17 . A system comprising:
a sensor configured to be at least partially disposed under a skin of a patient, the sensor configured to generate a signal; and
at least one processor configured to be operably connected with the sensor, the processor configured to:
evaluate a plurality of subsets of a calibration data set to determine a best subset;
determine a boundary associated with the best subset, wherein the boundary associated with the best subset is determined after the sensor is disposed at least partially under the skin of the patient;
determine at least one value outside the boundary as a possible outlier;
receive input from the patient concerning the possible outlier, the input comprising at least one of: meal intake, insulin administration, or exercise;
determine outlier information regarding the possible outlier based on the input received from the patient;
transform the calibration data set responsive to the outlier information by removing the possible outlier in response to the determination of the outlier information;
calculate a value representative of an analyte concentration based on the transformed calibration data set; and
transmit the value representative of the analyte concentration to a display.
18 . The system of claim 17 , wherein the processing, by the at least one processor, of the calibration data set responsive to the outlier information comprises temporarily or permanently removing the possible outlier from the calibration data set.
19 . The system of claim 17 , wherein the processing, by the at least one processor, of the calibration data set responsive to the outlier information comprises flagging an outlier and keeping the outlier in the calibration data set until a next data point is collected.