Flexible sensors and machine learning for heart monitoring

Abstract

Cardiovascular disease is the leading cause of death worldwide. Continuous heart monitoring is an effective approach in detecting irregular heartbeats and providing early warnings to patients. However, traditional cardiac monitoring systems have rigid interfaces and multiple wiring components that cause discomfort when continuously monitoring the patient long-term. To address those issues, flexible and comfortable sensing devices are critically needed, and they could also better match the dynamic mechanical properties of the epidermis to collect accurate cardiac signals. In this review, we discuss the principles of the major mechanisms of heart monitoring approaches as well as traditional cardiovascular monitoring devices. Based on key challenges and limitations, we propose design principles for flexible cardiac sensing devices. Recent progress of cardiac sensors based on various nanomaterials and structural designs are closely reviewed, along with the fabrication methods utilized. Moreover, recent advances in machine learning have significantly implemented a new sensing platform for the multifaceted assessment of heart status, and thus is further reviewed and discussed. Such strategies for designing flexible sensors and implementing machine learning provide a promising means of automatically detecting real-time cardiac abnormalities with limited or no human supervision while comfortably and continuously monitoring the patient’s cardiac health.