Falls classification using tri-axial accelerometers during the five-times-sit-to-stand test.

Area of Science:

• Gerontology
• Biomechanics
• Medical Technology

Background:

• The five-times-sit-to-stand test (FTSS) is a common clinical tool for assessing lower limb strength and falls risk.
• Current FTSS assessment relies on timing, which may not fully capture the nuances of movement related to falls risk.

Purpose of the Study:

• To investigate the utility of tri-axial accelerometers in quantifying movement during the FTSS for a more objective falls risk assessment.
• To develop and validate a model using accelerometer-derived features to classify older adults based on their falls history.

Main Methods:

• 39 older adults (19 with a history of falls) performed the FTSS at home with accelerometers on the thigh and sternum.
• Analysis focused on mean and variation of acceleration, jerk, and spectral edge frequency during sit-stand-sit phases.
• A classification model was built using reliable features (ICC > 0.7) via sequential forward feature selection and cross-validation.

Main Results:

• A model using four accelerometer-derived features achieved 74.4% classification accuracy for falls status.
• This accelerometer-based model demonstrated higher performance (80.0% specificity, 68.7% sensitivity) compared to using FTSS time alone.
• Key features included variations in acceleration, jerk, and spectral edge frequency.

Conclusions:

• Accelerometer-based quantification of FTSS movement provides a robust and accurate method for assessing falls risk in older adults.
• This technology offers a more objective and potentially more sensitive alternative to traditional timed assessments.
• Further development could enhance clinical decision-making and personalized fall prevention strategies.