Assessing real-world gait with digital technology? Validation, insights and recommendations from the Mobilise-D consortium
Digital mobility outcomes (DMOs) derived from wearable data require technical validation. This study comparatively assessed DMOs estimated from real-world gait data across six cohorts, focusing on gait sequence detection, initial contact detection (ICD), cadence (CAD), and stride length (SL). Twenty healthy older adults, 20 people with Parkinson’s disease, 20 with multiple sclerosis, 19 with proximal femoral fracture, 17 with chronic obstructive pulmonary disease, and 12 with congestive heart failure were monitored for 2.5 h in real-world conditions using a single lower-back wearable. A reference system combining inertial sensors, distance sensors, and pressure insoles was used for comparison. Three gait sequence detection, four ICD, three CAD, and four SL algorithms were evaluated using accuracy, sensitivity, specificity, predictive values, and errors. The effects of walking bout (WB) duration and speed were also analyzed. Two cohort-specific top-performing algorithms were identified for gait sequence detection and CAD, and a single best algorithm for ICD and SL. Gait sequence detection showed good performance (sensitivity > 0.73, specificity > 0.95, accuracy > 0.94). ICD and CAD showed excellent results (sensitivity > 0.79, predictive values > 0.89, relative error < 11% for ICD and < 8.5% for CAD). SL performance was lower (absolute error < 0.21 m). The proximal femoral fracture cohort showed the lowest performance. Short WBs and slow gait speeds (< 0.5 m/s) reduced performance, particularly for CAD and SL. The algorithms enabled robust estimation of key DMOs. Algorithm choice for gait sequence detection and CAD should be cohort-specific, especially for slow walkers and those with gait impairments. Short WBs and slow walking speed negatively affected performance. Trial registration: ISRCTN-12246987.
Supplementary resources
Preview
Rights
Use and reproduction:
Please note that individual components of the publication may be subject to other licensing or copyright conditions.