Abstract
<jats:p>Wearable glucose monitoring demands ultra-low-power local processing, but conventional neural networks rely on energy-intensive multiply–accumulate (MAC) operations that limit battery life. This study shows that a Spiking Neural Network (SNN), built on a regression adapted Leaky Integrate-and-Fire (LIF) neuron, can estimate blood glucose from multi-frequency bioimpedance and auxiliary biosignals with clinical-grade accuracy at a fraction of the computational cost. Using data from 98 patients (717 measurements, eGluco3 device, Azambuja Hospital, Brusque, Brazil) evaluated by 5-fold walk-forward cross-validation under ISO 15197:2013, three main findings emerge. First, a new calibration method—the Patient Fingerprint, built from each patient’s first K sensor readings—outperforms conventional one-hot patient encoding (14.2 ± 2.6 mg/dL vs. 15.4 ± 3.3 mg/dL mean absolute error) while working for patients never seen during training, a capability one-hot encoding lacks entirely. Second, this fingerprint model reaches 100% of samples within Consensus Error Grid Zones A+B across all validation folds, meeting the clinical-safety threshold, and does so without requiring any demographic or clinical metadata—sensor history alone renders such records redundant. Third, replacing the analog input encoding with a multiplication-free rate-coding scheme removes all first-layer MAC operations at a cost of only 2.7 mg/dL additional error, defining a concrete, quantified accuracy–hardware trade-off for neuromorphic deployment. Together, these results demonstrate that SNNs offer a clinically validated, calibration-free, and computationally efficient path to continuous glucose estimation on embedded wearable devices.</jats:p>