Situation Recognition and Medical Data Analysis in Pervasive Health Environments

Research Article

Operationalizing a wireless wearable fall detection sensor for older adults

Download622 downloads
  • @INPROCEEDINGS{10.4108/icst.pervasivehealth.2012.248643,
        author={Tigest Tamrat and Stan Kachnowski and Sonia Rupcic and Margaret Griffin and Tom Taylor Taylor and James Barfield},
        title={Operationalizing a wireless wearable fall detection sensor for older adults},
        proceedings={Situation Recognition and Medical Data Analysis in Pervasive Health Environments},
        publisher={IEEE},
        proceedings_a={PERVASENSE},
        year={2012},
        month={7},
        keywords={fall detection; actigraphy; activity monitors; elderly populations},
        doi={10.4108/icst.pervasivehealth.2012.248643}
    }
    
  • Tigest Tamrat
    Stan Kachnowski
    Sonia Rupcic
    Margaret Griffin
    Tom Taylor Taylor
    James Barfield
    Year: 2012
    Operationalizing a wireless wearable fall detection sensor for older adults
    PERVASENSE
    IEEE
    DOI: 10.4108/icst.pervasivehealth.2012.248643
Tigest Tamrat1, Stan Kachnowski2,*, Sonia Rupcic1, Margaret Griffin1, Tom Taylor Taylor3, James Barfield3
  • 1: Healthcare Innovation and Technology Lab
  • 2: Healthcare Innovation & Technology Lab; Indian Institute of Technology-New Delhi
  • 3: Lifecomm, LLC
*Contact email: swk16@hitlab.org

Abstract

Falls are the leading cause of disability and injury- related deaths among older adults, resulting in over 1.6 million annual emergency hospitalizations in the United States. Fall detection devices often rely on dramatized falls when developing algorithms. This study used tri-axial accelerometers worn by older adult research subjects in order to (1) collect false positive data (2) capture potential fall events and (3) evaluate the usability of the device among this target population. Twelve older adults wore activity monitors while participating in structured and unstructured activities. The study collected data on 120 patient days, yielding 492.5 hours of monitored time. Actigraphy data of annotated activities were used to define parameters for refining the algorithm. No falls occurred during the study, but valuable false positive data were collected. The study also obtained information on the usability of the devices and revealed user perspectives on commercializing the final product.