Rethinking Inexpensive Wearables in the Era of AI: From Motion Analytics to Mobile Health

Since their emergence, wearables have been widely proposed and utilized for counting steps, monitoring heart rate, and enhancing user interactions, suggesting their growing importance in both research and commercial domains. However, despite their ubiquity, the application scope of current wearables remains limited, largely due to an insufficient understanding of sensor signals. Mainstream devices such as smartphones and smartwatches predominantly support basic functionalities like human activity recognition and indoor navigation. However, enabling more advanced applications necessitates a deeper understanding of sensor signals within these wearables. Without such insight, these applications often depend on specialized, high-end hardware or fail to generalize across diverse conditions. For instance, sign language recognition, which requires fine-grained finger motion tracking, typically depends on bulky sensor gloves, vision-based systems that are expensive and intrusive, or wireless signal based systems that require specialized hardware, making them unsuitable for daily use. Similarly, while current smartwatches/rings can estimate heart rate, advanced health markers such as cardiac output, which offer insights into cardiac dysfunction, acute stress responses, and cognitive decline, are limited to specialized devices that cost thousands of US dollars.