Algorithms to Improve Exercise Intensity Estimation by Using Wrist- and Hip-Worn Accelerometers and Treadmill Exercise

Health management and physical rehabilitation often involve personalized plans. Exercise intensity measurements help tailor these plans to the specific needs, abilities, and medical conditions of the individual. Accelerometers can be utilized for estimating exercise intensity, although their accuracy may be somewhat limited. This study’s focus was to determine whether accelerometers with the assistance of specialized algorithms could replace conventional heart rate monitoring approaches for exercise intensity estimations. Seventeen participants were enrolled and performed a discontinuous exercise test on a treadmill with seven different exercise regimens –three walking and four running routines. Acceleration measurements were acquired using two tri-axial accelerometers worn at the hip and wrist. Pulse measurements were simultaneously collected to calculate the heart rate reserve. The measurements acquired from hip-worn and wrist-worn accelerometers were processed using integral, peak, and mean amplitude deviation algorithms and correlated with the different walking and running speeds. Strong correlations between resultant accelerations and walking speeds (
$r =$
ׅn913,
$P \lt $
ׅ001) were observed in the data collected by the combination of a hip-worn accelerometer and the mean amplitude deviation algorithm. Strong correlations for running speeds (
$r =$
ׅ786,
$P \lt $
ׅ001) were also observed in the data collected by the combination of a wrist-worn accelerometer and the integral algorithm. The heart rate reserve increased with incremental speeds (walking:
$r =$
ׅ850,
$P \lt $
ׅ001; running:
$r =$
ׅ764,
$P \lt $
ׅ001). The appropriate combination of accelerometer placement and estimation algorithm provides excellent exercise intensity estimations without the limitations presented by heart rate index referencing.