Activity-Aware Vital SignMonitoring Based on a Multi-Agent Architecture.

Area of Science:

• Biomedical Engineering
• Health Informatics
• Wearable Technology

Background:

• Remote vital sign monitoring using wearable sensors offers continuous patient assessment outside clinical settings.
• Activity-dependent vital sign variations necessitate dynamic monitoring solutions for accurate health evaluation.
• Automatic alerts for health deterioration enhance patient care and timely intervention.

Purpose of the Study:

• To propose a multi-agent system architecture for dynamic vital sign monitoring during daily physical activities.
• To accurately classify activities of daily living using wearable sensor data.
• To develop a method for automatically extracting vital sign threshold ranges for different physical activities to support remote health evaluation.

Main Methods:

• Utilized a multi-agent paradigm with specialized agents processing signals from chest, wrist, and ankle accelerometer sensors.
• Employed ontology-based models to manage data heterogeneity from multiple wearable sensor sources.
• Validated the system using a real-life dataset with subjects performing various physical activities.

Main Results:

• Achieved 95.25% accuracy in intersubject activity classification.
• Demonstrated promising results in remote health status evaluation through automatically extracted vital sign threshold ranges.
• Successfully integrated and processed data from multiple wearable sensors for enhanced performance.

Conclusions:

• The proposed multi-agent system effectively monitors vital signs during physical activities, adapting to varying intensities.
• Ontology-based data integration and multi-sensor fusion improve the accuracy and reliability of remote health monitoring.
• The system shows significant potential for proactive and personalized remote patient care.