Whole-cycle and time-specific validation of a GUI-based ground reaction force estimation tool for clinical gait analysis without a force plate

Relying on patients to make clean contacts with multiple force plates during clinical gait analysis can be time-consuming and dissuade clinicians from collecting biomechanical data. Several methods have been proposed to estimate ground reaction forces (GRFs) without force plates, though many can be computationally intensive. As an alternative to measuring GRFs, we evaluated the whole cycle and time frame accuracy of a built-in, GUI-based GRF estimation tool. Twenty-seven healthy adult participants walked over two consecutive force plates. We evaluated the accuracy of estimated GRFs against force plates as the gold standard using correlation, residual error analysis, and statistical parametric mapping (SPM). Estimated GRF magnitudes were accurate in the anterior-posterior (R²=0.86, %RMSE=7.28), and vertical directions (R²=0.86, %RMSE=6.09), but not in the medio-lateral direction (R²=0.30, %RMSE=12.46). SPM revealed errors in the sagittal plane in mid- and late-stance, with no specific errors in the medio-lateral direction. Estimated centers of pressure (CoPs) showed good accuracy (R²≥0.99, %RMSE≤4.46). However, SPM showed estimated CoPs were consistently anterior to the true CoPs. Sagittal GRFs are predicted with good accuracy, showing whole cycle errors comparable to more complex methods, while also displaying little early stance error and offering user-friendly implementation.