Toward real-time accurate fall/fall recovery detection system by incorporating activity information

• Published in: Proceedings of 2012 IEEE-EMBS International Conference on Biomedical and Health Informatics pp. 196 - 199
• Main Authors: Pannurat, N., Theekakul, P., Thiemjarus, S., Nantajeewarawat, E.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.01.2012
• ISBN: 9781457721762; 1457721767
• ISSN: 2168-2194; 2168-2208
• DOI: 10.1109/BHI.2012.6211543

This study presents a detailed summary of automatic fall detection system based on wearable sensor(s) and a real-time fall detection system prototype developed based on Java Expert System Shell (JESS), featuring the fall, fall recovery, fall direction and activity status before/after fall. Through the rule sets in the knowledge base, we illustrate how the activity information can be used to indicate the fall recovery and fall direction, as well as enhancing the accuracy of fall detection. The system has been validated against 13 types of falls and 12 ADLs acquired from 12 subjects.