Abstract
<jats:p>Continuous glucose monitoring (CGM) has improved diabetes care, yet cost, wear burden, and reduced accuracy in the hypoglycemic range still motivate complementary, low-burden sensing. Consumer wearables that record electrocardiography (ECG), photoplethysmography (PPG), electrodermal activity (EDA), and related signals are widely available, and a growing literature maps them to glucose values, glycemic curves, or hypo-/hyperglycemia risk. Progress remains uneven: free-living signals are unstable, physiology-to-glucose mappings are indirect and non-deterministic, public multimodal datasets are small, evaluation protocols are heterogeneous, and cross-subject generalization is limited. Based on representative works in this field, this preprint contributes: (1) a three-task taxonomy (risk, value, curve) and how each aligns with wearable modalities; (2) a six-lineage evidence map with a qualitative exposure table of typical protocols and reporting gaps, treating CGM-only forecasting as a Task-A contrast baseline; (3) a synthesis of quality, uncertainty, and interpretability gaps plus a literature-derived minimal reporting checklist; and (4) an open-problem agenda. These deliverables clarify applicability and trust boundaries for wearable physiology as a glycemic decision aid under declared conditions, rather than as an undeclared CGM replacement.</jats:p>