Using Smartphones to Track Gait Events: A Neural Network Approach

Trying to understand how you walk with just your smartphone. Researchers did exactly that! They created a clever algorithm using neural networks to detect when your foot hits the ground (heel strike) and when it lifts off (toe-off) while carrying a phone. Fifty-two adults walked on a treadmill with phones in different pockets and at various speeds. An advanced system tracked their leg movements to find the exact moments of these gait events. The phone's sensors sent data in 20-ms chunks, which the algorithm classified as either a heel strike or toe-off for each leg. With 80% of the data used for training and 20% for testing, the algorithm was amazingly accurate—92% overall! For the right leg, it was even better, around 94%, but slightly lower for the left, at about 91%. The timing error was less than 3% of a gait cycle, which is super-fast and accurate. This study is a big step towards using smartphones for gait analysis without needing special sensors, but more work is needed to make it work perfectly in the real world.