How should we interpret retrospective blood glucose measurements? Sampling and Interpolation

摘要

This study investigates blood glucose (BG) measurement interpolation techniques to represent intermediate BG dynamics, and the effect resampling of retrospective BG data has on key glycemic control (GC) performance results. Many GC protocols in the ICU have varying BG measurement intervals with gaps ranging from 0.5 to 4 hrs. Sparse data poses problems in model fitting techniques and GC performance comparisons, and thus interpolation is required to assume a continuous solution. Retrospective data from SPRINT in the Christchurch Hospital Intensive Care Unit (ICU) (2005-2007) was used to analyze various interpolation techniques. Piece-wise linear, spline and cubic interpolation functions, which force lines through data, as well as 1st and 2nd Order B-spline basis functions, used to identify the data, are investigated. Dense data was thinned to increase sparsity and obtain measurements (Hidden measurements) for comparison after interpolation. All of the piece-wise functions performed considerably better than the fitted interpolation functions. Linear piece-wise interpolation performed the best having a mean RMSE 0.39 mmol/L, within 2 standard deviations of the BG sensor error. The effect of minutely vs hourly sampling of the interpolated trace on key GC performance statistics was investigated using the retrospective data received from STAR GC in the Christchurch Hospital Intensive Care Unit (ICU), New Zealand (2011-2015). Minutely sampled BG resulted in significantly different key GC performance when compared to raw sparse BG measurements. Linear piece-wise interpolation provides the best estimate of intermediate BG dynamics and all analyses comparing GC protocol performance should use minutely linearly interpolated BG data.