Automatic identification of gait events during walking on uneven surfaces.

The accurate detection of gait events is essential for clinical gait analysis. Aside from speed, surface characteristics like planarity and compliance can affect gait kinematics. Therefore detection of kinematic gait events on uneven surfaces may be inaccurate. To date, no study has investigated the possible influence of surface characteristics on gait event detection. Thus, the purpose of this study was to assess and compare the performance of four kinematic-based gait event detection algorithms (horizontal heel-heel displacement, foot velocity, heel/toe-PSIS displacement, peak hip extension) during walking on three surfaces with different degrees of planarity. Kinematic and force plate data were collected on thirteen athletes during two self-selected walking speeds at a normal (1.30±0.03m/s) and fast pace (1.70±0.10m/s). Footstrike and toe-off events were calculated by the algorithms and compared to vertical ground reaction force as a reference. The main findings of the study were: (1) surface configuration had an effect on algorithm accuracy (p<0.010, 0.84<d<2.79); (2) the vertical foot-velocity profile provided the lowest RMSE for footstrike (8.8-14.6ms) during normal walking and toe-off (15.4-24.9ms) during normal and fast walking on all surfaces; (3) horizontal heel-ankle displacement provided the lowest RMSE for footstrike during fast walking on all surfaces (RMSE: 8.9-13.8ms). Overall, the vertical foot-velocity algorithm provided low RMSE for all conditions, is easy to apply and thus recommended for gait event detection.

RCT Entities:   Population Interventions Outcomes