Abstract
Children with cerebral palsy may have difficulty walking and may fall frequently, resulting in a decrease in their participation in school and community activities. It is desirable to assess the effectiveness of mobility therapies for these children on their functioning during everyday living. Over 50 hours of tri-axial accelerometer and digital video recordings from 35 children with cerebral palsy and 51 typically-developing children were analyzed to develop algorithms for automatic real-time processing of the accelerometer signals to monitor a child's level of activity and to detect falls. The present fall-detection algorithm has 100% specificity and a sensitivity of 100% for falls involving trunk rotation. Sensitivities for drops to the knees and to the bottom are 72% and 78%, respectively. The activity and fall-detection algorithms were implemented in a miniature, battery-powered microcontroller-based activity/fall monitor that the child wears in a small fanny pack during everyday living. The monitor continuously logs 1-min. activity levels and the occurrence and characteristics of each fall for two-week recording sessions. Pre-therapy and post-therapy recordings from these monitors will be used to assess the efficacies of alternative treatments for gait abnormalities.