An efficient user-customisable multiresolution classifier fall detection and diagnostic system

Ryan M. Gibson, Abbes Amira, Pablo Casaseca-de-la-Higuera, Naeem Ramzan, Zeeshan Pervez

Research output: Chapter in Book/Report/Conference proceedingConference contribution

3 Citations (Scopus)


Falling can cause significant injury, where quick medical response and fall information are critical to providing aid. In this paper we present a wearable wireless fall detection system utilising a Shimmer accelerometer device, where important additional information is obtained, such as direction and strength of the occurred fall instance. Discrete Wavelet Transforms and multiresolution wavelet analysis are used to accurately determine fall occurrence and additionally determine the strength of the fall. The wavelet signal is additionally evaluated with Principal Component Analysis to generate a decision tree classifier for fall occurrence, strength and direction. Test subjects undertook fall and Activities of Daily Living experiments to generate data for wavelet and Principal Component Analysis. The presented fall detection and diagnostic system obtained highly accurate and robust fall detection with both methods, while the decision tree strength analysis demonstrated a better fall strength response.
Original languageEnglish
Title of host publicationIEEE 26th International Conference on Microelectronics (ICM 2014)
ISBN (Electronic)9781479981533
Publication statusPublished - 2 Apr 2015


  • decision trees
  • principal component analysis
  • wavelet domain
  • acceleration
  • accuracy
  • wavelet analysis
  • signal resolution


Dive into the research topics of 'An efficient user-customisable multiresolution classifier fall detection and diagnostic system'. Together they form a unique fingerprint.

Cite this