Adaptive calibration algorithm for plasma glucose estimation in continuous glucose monitoring.

Minimally or noninvasive continuous glucose monitors estimate plasma glucose from compartments alternative to blood, and may revolutionize the management of diabetes. However, the accuracy of current devices is still poor and it may partly depend on low performance of the implemented calibration algorithm. Here, a new adaptive calibration algorithm based on a population local-model-based intercompartmental glucose dynamic model is proposed. The novelty consists in the adaptation of data normalization parameters in real time to estimate and compensate patient's sensitivity variations. Adaptation is performed to minimize mean absolute relative deviation at the calibration points with a time window forgetting strategy. Four calibrations are used: preprandial and 1.5 h postprandial at two different meals. Two databases are used for validation: 1) a 9-h CGMS Gold (Medtronic, Northridge, USA) time series with paired reference glucose values from a clinical study in 17 subjects with type 1 diabetes; 2) data from 30 virtual patients (UVa simulator, Virginia, USA), where inter- and intrasubject variability of sensor's sensitivity were simulated. Results show how the adaptation of the normalization parameters improves the performance of the calibration algorithm since it counteracts sensor sensitivity variations. This improvement is more evident in one-week simulations.

A multiple local models approach to accuracy improvement in continuous glucose monitoring.

BACKGROUND: Continuous glucose monitoring (CGM) devices estimate plasma glucose (PG) from measurements in compartments alternative to blood. The accuracy of currently available CGM is yet unsatisfactory and may depend on the implemented calibration algorithms, which do not compensate adequately for the differences of glucose dynamics between the compartments. Here we propose and validate an innovative calibration algorithm for the improvement of CGM performance. METHODS: CGM data from GlucoDay(®) (A. Menarini, Florence, Italy) and paired reference PG have been obtained from eight subjects without diabetes during eu-, hypo-, and hyperglycemic hyperinsulinemic clamps. A calibration algorithm based on a dynamic global model (GM) of the relationship between PG and CGM in the interstitial space has been obtained. The GM is composed by independent local models (LMs) weighted and added. LMs are defined by a combination of inputs from the CGM and by a validity function, so that each LM represents to a variable extent a different metabolic condition and/or sensor-subject interaction. The inputs best suited for glucose estimation were the sensor current I and glucose estimations G, at different time instants [I(k), I(k)(-1), G(k)(-1)] (IIG). In addition to IIG, other inputs have been used to obtain the GM, achieving different configurations of the calibration algorithm. RESULTS: Even in its simplest configuration considering only IIG, the new calibration algorithm improved the accuracy of the estimations compared with the manufacturer's estimate: mean absolute relative difference (MARD)=10.8±1.5% versus 14.7±5.4%, respectively (P=0.012, by analysis of variance). When additional exogenous signals were considered, the MARD improved further (7.8±2.6%, P<0.05). CONCLUSIONS: The LM technique allows for the identification of intercompartmental glucose dynamics. Inclusion of these dynamics into the calibration algorithm improves the accuracy of PG estimations.