Invited Paper: Feature-to-Classifier Co-Design for Mixed-Signal Smart Flexible Wearables for Healthcare at the Extreme Edge

TL;DR

This paper introduces a holistic co-design framework for flexible wearable healthcare devices, integrating analog feature extraction and ML-based classifier optimization to achieve ultra-efficient, low-power, and accurate systems suitable for disposable applications.

Contribution

It presents the first analog feature extractors in flexible electronics and a hardware-aware NAS-inspired feature selection strategy for system-wide optimization.

Findings

Achieves highly accurate healthcare monitoring with ultra-area efficiency.

Reduces hardware cost of feature extraction and ADCs significantly.

Demonstrates suitability for disposable, low-power wearable systems.

Abstract

Flexible Electronics (FE) offer a promising alternative to rigid silicon-based hardware for wearable healthcare devices, enabling lightweight, conformable, and low-cost systems. However, their limited integration density and large feature sizes impose strict area and power constraints, making ML-based healthcare systems-integrating analog frontend, feature extraction and classifier-particularly challenging. Existing FE solutions often neglect potential system-wide solutions and focus on the classifier, overlooking the substantial hardware cost of feature extraction and Analog-to-Digital Converters (ADCs)-both major contributors to area and power consumption. In this work, we present a holistic mixed-signal feature-to-classifier co-design framework for flexible smart wearable systems. To the best of our knowledge, we design the first analog feature extractors in FE, significantly reducing feature extraction cost. We further propose an hardware-aware NAS-inspired feature selection strategy within ML training, enabling efficient, application-specific designs. Our evaluation on healthcare benchmarks shows our approach delivers highly accurate, ultra-area-efficient flexible systems-ideal for disposable, low-power wearable monitoring.