Biomedical Signal Processing: The Cornerstone of Artificial Intelligence in Healthcare Wearables

Abstract

Health sensors and remote measurement tools have saved lives through the possibility of continuous monitoring and intervention tools, and over the years their use has expanded to non-medical areas such as fitness and perceived well-being. This expansion has led to unprecedented data collection, especially since biomedical sensors are now ubiquitous in everyday devices such as smartwatches and smartphones. While these devices can be disruptive research tools and even clinical tools, they pose technological and socio-economic challenges that can limit their impact. Here, we highlight these challenges, including the use of proxies for clinical reference measurements, uncertainties resulting from the presence of noise, complexity of physiological systems, and statistical methods used for data interpretation.