Detecting tap events from a foot-tapping test with an accelerometer: Validation of Teager-Kaiser energy operator and discrete wavelet transform

Foot-tapping is used to evaluate functional and cognitive ability of people with and without neurological disorders. Accurate tap event detection during this test is key. This study validated the accuracy of two algorithms: Ⅰ) Teager-Kaiser energy operator in conjunction with the Maximal Overlap Discrete Wavelet Transform (TKEO-MODWT) and II) Discrete Wavelet Transform (DWT), for detecting tap events from accelerometer data during a unilateral foot-tapping test. These algorithms have been validated for tap detection during finger-tapping tests in healthy younger adults and people with Parkinson’s disease. However, their accuracy for detecting taps during a foot-tapping test has not been explored. Thirty-three participants (15 older adults aged 65 and above, and 18 young-midlife adults 64 years old and under) performed a dominant and nondominant unilateral foot-tapping test. Data were recorded using an accelerometer and a force plate. The sensitivity, precision, mean absolute error in tap detection, and timing detected by the two algorithms from the accelerometer data were compared against true taps recorded using a force plate. Both algorithms (TKEO-MODWT and DWT) were robust for detecting foot-tapping taps. Algorithm Ⅰ (TKEO-MODWT) consistently outperformed algorithm Ⅱ (DWT), achieving higher sensitivity and precision. Additionally, algorithm Ⅰ (TKEO- MODWT) resulted in lower mean absolute error between detected taps and the true taps for both feet in both groups. TKEO-MODWT provides an accurate tool for objectively assessing unilateral foot-tapping in young-midlife adults and older adults using only an accelerometer, and could add to the clinical assessment of functional and cognitive assessment of older adults.