Classification of daily physical activities from a single kinematic sensor

Area of Science:

• Biomedical Engineering
• Health Informatics
• Wearable Technology

Background:

• Monitoring physical activity and autonomy is crucial for supervising elderly or fragile individuals at home.
• Assessing time spent in different postures (lying, sitting, standing), walking periods, and postural transitions provides key health indicators.
• Existing methods may lack the unobtrusiveness or accuracy needed for continuous home-based health monitoring.

Purpose of the Study:

• To develop and validate algorithms for detecting physical activities and postural transitions using a single, chest-mounted accelerometer.
• To enable continuous, unobtrusive monitoring of elderly individuals' activity levels within a Health Smart Home framework.
• To provide objective data on patient activity and autonomy for health evaluation.

Main Methods:

• Development of distinct algorithms for posture detection, walking detection, and postural transition detection (sit-to-stand, back-to-sit).
• Utilizing a unique sensor comprising three accelerometers, worn on the chest.
• Validation of algorithms using real-world data collected from elderly subjects performing daily activities.

Main Results:

• The developed algorithms successfully detected various physical activities, including lying, sitting, standing, and walking.
• Accurate identification of postural transitions such as sit-to-stand and back-to-sit was achieved.
• Experimental results demonstrated the feasibility of using the chest-worn sensor for monitoring elderly subjects' daily activities.

Conclusions:

• A novel method using a single chest-worn accelerometer sensor can effectively monitor physical activities and postural transitions in elderly individuals.
• This technology holds promise for enhancing the supervision and health assessment of elderly people in home environments.
• The developed algorithms provide a foundation for future advancements in smart home healthcare solutions.