Automated Detection of Gait Events and Travel Distance Using Waist-worn Accelerometers Across a Typical Range of Walking and Running Speeds

We investigate the use of a single smartphone-based accelerometer placed near the body's center of mass to extract temporospatial gait clinical features (CFs) in children with Duchenne muscular dystrophy (DMD) and typically developing (TD) peers during ambulation at varying gait speeds. Machine learning-based algorithms (ML) were developed to estimate temporospatial gait CFs, including the number of steps, distance, duration of the gait cycle, gait speed, and step length. The estimated temporospatial gait CFs were compared with ground-truth CFs determined by expert observers during clinical testing using the 6-minute walk test, 100-meter run/walk, and self-selected free walk. Additionally, the extracted CFs from the accelerometer were compared with CFs extracted by the pedometer feature built into the smartphone. The study findings indicate that a single accelerometer placed near the body's center of mass can accurately measure CFs across different gait speeds in both TD and DMD peers, suggesting the potential for measuring CFs in the community without the need for ground reaction force (GRF) measurements.