Improving the Robustness of the Glycemic Variability Percentage Metric to Sensor Dropouts in Continuous Glucose Monitor Data
Continuous glucose monitors generate significant volumes of high frequency blood glucose data. Analysis of this data by a physician may entail the calculation of various glycemic variability metrics. In this paper, we consider the problem of metric robustness to sensor dropouts. We show that the standard metrics for glycemic variability are unreliable with missing data. A more recent metric, glycemic variability percentage, is shown to consistently underestimate glycemic variability as the amount of missing data increases. We therefore propose a new algorithm based on random sampling combined with linear regression to correct this underestimation, and show that the metric’s accuracy is significantly increased with our correction.
- 3.Cox, D.J., Gonder-Frederick, L., Ritterband, L., Clarke, W., Kovatchev, B.P.: Prediction of severe hypoglycemia. Diabetes Care 30(6), 1370–1373 (2007). https://doi.org/10.2337/dc06-1386. http://care.diabetesjournals.org/content/30/6/1370CrossRefGoogle Scholar
- 4.Danne, T., et al.: International consensus on use of continuous glucose monitoring. Diabetes Care 40(12), 1631–1640 (2017). https://doi.org/10.2337/dc17-1600. http://care.diabetesjournals.org/content/40/12/1631CrossRefGoogle Scholar
- 6.Hirsch, I.B., Balo, A.K., Sayer, K., Garcia, A., Buckingham, B.A., Peyser, T.A.: A simple composite metric for the assessment of glycemic status from continuous glucose monitoring data: implications for clinical practice and the artificial pancreas. Diabetes Technol. Ther. 19(S3), S-38–S-48 (2017). https://doi.org/10.1089/dia.2017.0080CrossRefGoogle Scholar
- 11.The Nightscout Foundation: About the Nightscout Data Commons on Open Humans (2014). http://www.nightscoutfoundation.org/data-commons/