Interstitial fluid glucose time-lag correction for real-time continuous glucose monitoring

Time lag between subcutaneous interstitial fluid and plasma glucose decreases the accuracy of real-time continuous glucose monitors. However, inverse filters can be designed to correct time lag and attenuate noise enabling the blood–glucose profile to be reconstructed in real time from continuous measurements of the interstitial-fluid glucose. We designed and tested a Wiener filter using a set of 20 sensor-glucose tracings (∼30 h each) with a 1-min sample interval. Delays of 10 ± 2 min (mean ± SD) were introduced into each signal with additive Gaussian white noise (SNR = 40 dB). Performance of the filter was compared to conventional causal and non-causal seventh-order finite-impulse response (FIR) filters. Time lags introduced an error of 5.3 ± 2.7%. The error increased in the presence of noise (to 5.7 ± 2.6%) and attempts to remove the noise with conventional low-pass filtering increased the error still further (to 7.0 ± 3.5%). In contrast, the Wiener filter decreased the error attributed to time delay by ∼50% in the presence of noise (from 5.7% to 2.60 ± 1.26%) and by ∼75% in the absence of noise (5.3% to 1.3 ± 1%). Introducing time-lag correction without increasing sensitivity to noise can increase CGM accuracy.

► CGM devices that measure ISF glucose, and are calibrated with capillary blood–glucose meters in order to estimate blood–glucose levels, have decreased accuracy due to the lag time between plasma glucose and glucose measured in the interstitial space. ► A first-order time lag is a reasonable model to explain the diffusion process. ► Time-lag correction of 10 min produced optimal results when applied to signals collected from 10 patients. ► Wiener filter enables patients to monitor their blood–glucose levels in real time. ► Real-time readings from current CGM devices lag blood glucose by the physiological delay between glucose measured in plasma and glucose measured in ISF plus FIR filter group delays. ► Wiener filter provides faster event detection and increased patient safety.