10th EAI International Conference on Body Area Networks

Research Article

A smartphone-centered wearable sensor network for fall risk assessment in the elderly

  • @INPROCEEDINGS{10.4108/eai.28-9-2015.2261433,
        author={Andrea Mannini and Angelo Sabatini},
        title={A smartphone-centered wearable sensor network for fall risk assessment in the elderly},
        proceedings={10th EAI International Conference on Body Area Networks},
        publisher={ACM},
        proceedings_a={BODYNETS},
        year={2015},
        month={12},
        keywords={wearable sensor networks fall risk assessment gait stability activity recognition},
        doi={10.4108/eai.28-9-2015.2261433}
    }
    
  • Andrea Mannini
    Angelo Sabatini
    Year: 2015
    A smartphone-centered wearable sensor network for fall risk assessment in the elderly
    BODYNETS
    ICST
    DOI: 10.4108/eai.28-9-2015.2261433
Andrea Mannini1,*, Angelo Sabatini1
  • 1: Scuola Superiore Sant'Anna
*Contact email: a.mannini@sssup.it

Abstract

Fall prevention is an important aspect to keep high the quality of life in aging. In this work, a wearable sensor network to automatically assess movement and its indicator for fall risk is proposed. The method is based on a smartphone linked to wearable devices. The proposed approach starts estimating the physical activity level using activity recognition algorithms running on the phone. Such estimation can be conducted by simply using the smartphone on its own. Activity recognition can also identify walking bouts to drive gait assessment tools using different sensing sources. In this regard a strategy to estimate stride-based gait stability indexes during ambulation was implemented. Waist and shank sensing nodes were involved in the computation of stride time, stride time variability, vertical acceleration variability and harmonic ratios. The proposed method can correctly recognize activity with a 95% accuracy within four classes. Ambulation in particular is recognized correctly in 97.2% of cases and gait stability indexes are then estimated capturing differences between controls and patients at high risk of falling.