Regularised Model Identification Improves Accuracy of Multisensor Systems for Noninvasive Continuous Glucose Monitoring in Diabetes Management
Figure 4
Representative recording sessions of Subjects AA03 (a) and AA05 (b). LASSO, Ridge, and EN models test over independent test data subset (continuous lines) versus RBG levels (open bullets). Bottom panels display two representative channels (no.2 and no.3 for subject on the left and on the right, resp.) entering the models, where occasional spikes and jumps are evident. MARD values for the experimental session on the right are of 53% (LASSO), 24.5% (Ridge), and 20.6% (EN), while for the experimental session on the left of 55.7% (LASSO), 34.8% (Ridge), and 34.7% (EN).
