Intermittent Neuromorphic Wearable Systems

Wearable medical devices have become a focal point of research and development, particularly for their use in long-term monitoring of patients with chronic diseases. These devices offer significant advantages, including early detection of complications, reduced hospital readmission rates, and overall lower healthcare costs by enabling a more proactive approach to patient care. In recent years, the integration of Artificial Intelligence (AI) into wearable systems has gained considerable attention, further enhancing the capabilities of these devices. However, AI-driven wearables consume significantly more power than their traditional counterparts, which can limit the device's lifetime. To alleviate this problem, this paper presents a novel framework combining two state-of-the-art techniques for reducing embedded AI energy consumption: neuromorphic computing and intermittent computing. As a proof of concept, our framework is applied to an Electromyography (EMG) application to classify various hand gestures using the Ninapro DB2 dataset. Our findings demonstrate that the proposed framework supports the creation of wearable AI devices with state-of-the-art energy efficiency.