Improving Accuracy and Precision of Glucose Sensor Profiles: Retrospective Fitting by Constrained Deconvolution

摘要: Frequent and accurate reference measurements of blood-glucose (BG) concentration are key for modeling computing outcome metrics in clinical trials but difficult, invasive, costly to collect. Continuous glucose monitoring (CGM) is a minimally-invasive technology that has the requested temporal resolution substitute BG references such scope, still lacks precision accuracy. In this paper, we propose an algorithm retrospectively reconstructs reliable continuous-time profile aforementioned purposes, by simultaneously exploiting high accuracy (possibly sparse) CGM data. The performs constrained semiblind deconvolution two steps: first, it estimates unknown parameters model accounting plasma-interstitum diffusion sensor inaccurate calibration; then, performing regularized data, subject additional constraint reconstructed lay within confidence interval available references. was tested on 24 datasets collected 20 h trial where records median 13 samples per day were available. Mean absolute relative deviation reduced (from 15.71% 8.84%) with respect unprocessed so did error evaluation outcomes (e.g., halved time-in-hypo assessment). profile, view its improved precision, suitable assessment, other offline applications.