Effects of digital filtering on peak acceleration and force measurements for artistic gymnastics skills

Rhiannon A. Campbell*, Elizabeth J. Bradshaw, Nick Ball, Adam Hunter, Wayne Spratford

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

12 Citations (Scopus)


Low-pass filters are ideal when filtering human movements, however the effectiveness of such filters relies on the correct selection of the cut-off frequency. The aim of this study was to determine the most appropriate filter cut-off for acceleration- and force-time data when measuring peak resultant acceleration (PRA) and ground reaction force (PRGRF) during gymnastics landings. Sixteen gymnasts executed backward handsprings and backward somersault landings onto a matted force plate while wearing four inertial measurement units (IMUs). Acceleration- and force-time data were filtered using a fourth-order Butterworth filter at different cut-off frequencies ranging from raw through to 250 Hz. Residual analysis plots were produced, and the PRGRF and PRA for all IMUs were calculated for each participant and skill at all cut-off frequencies. Descriptive statistics, model II linear regressions and Bland-Altman plots were conducted. Results indicated that a minimum 85 Hz cut-off is optimal. High cut-off frequencies (>80 Hz) showed good linear relationships and had minimal mean bias compared with raw values, indicating that either filtered (above ~85 Hz) or raw signals can be used. It is suggested that for applied sports settings no filtering is needed, however a minimum cut-off of 85 Hz should be implemented for research purposes.

Original languageEnglish
Pages (from-to)1859-1868
Number of pages10
JournalJournal of Sports Sciences
Issue number16
Publication statusPublished - 17 Aug 2020
Externally publishedYes


  • accelerometer
  • biomechanics
  • landing
  • load
  • Training


Dive into the research topics of 'Effects of digital filtering on peak acceleration and force measurements for artistic gymnastics skills'. Together they form a unique fingerprint.

Cite this