Power Optimized Smartwatch-Earbuds Multimodal System for Monitoring Activities of Daily Living

Area of Science:

• Biomedical Engineering
• Wearable Technology
• Health Informatics

Background:

• Monitoring activities of daily living (ADL) is crucial for managing chronic diseases and promoting healthy lifestyles.
• Existing systems often face challenges with power consumption and continuous passive monitoring.

Purpose of the Study:

• To develop a multi-modal, power-efficient, end-to-end mobile system for continuous, passive, and accurate detection of daily activity classes using inertial sensors from smartwatches and earbuds.
• To assess the system's impact on battery life and its potential clinical relevance for chronic disease patients.

Main Methods:

• Collected posture, stationary, and moving activity data from 40 diverse subjects using smartwatches and earbuds.
• Developed a novel power-optimized system including optimized sampling/transfer rates, data buffering, background services, and model size reduction.
• Evaluated system performance using macro recall score and compared battery consumption against a power-agnostic approach.

Main Results:

• Achieved a 93% macro recall score in detecting various daily activities.
• Demonstrated significant battery power reduction: 80% for smartphones, 40% for smartwatches, and 33.33% for earbuds compared to a power-agnostic system.
• Estimated continuous run times of 50 hours (smartphone), 16.67 hours (smartwatch), and 25 hours (earbuds) without recharging.

Conclusions:

• The developed end-to-end power-optimized activity detection system enables continuous, passive lifestyle monitoring.
• This system offers substantial battery savings, enhancing usability for long-term data collection.
• Clinical relevance lies in assisting healthcare providers to monitor lifestyle and sedentary behavior in chronic disease patients for improved treatment.