Cross-Validation Of Cadence As A Predictor For The Walk-To-Run Transition In Healthy Adults

BACKGROUND: The walk-to-run transition (WRT) is the natural yet sudden shift from walking to running. A prior study reported that cadence was a stronger predictor of the WRT than other commonly used indicators such as gait speed of ~2.1 m/s and a Froude number (a dimensionless value that accounts for leg length) of 0.5 (unitless). Specifically, cadence values of 135 and 140 steps/min demonstrated excellent classification accuracy (>90%) for WRT prediction in healthy adults. Notably, the authors called for cross-validation in a separate sample to confirm their findings. PURPOSE: To determine the sensitivity, specificity, and accuracy of cadence-based thresholds (135 and 140 steps/min) for predicting the WRT in a secondary, independent sample. METHODS: Participants (N = 44, aged 22 ± 3.5 years, 68% women) completed four*3-min treadmill bouts at 0.89, 1.34, 1.78 and 2.24 m/s (2, 3, 4, and 5 mph). Cadence was calculated by dividing the total number of steps (via hand count and video) by bout duration. Each treadmill bout was classified as “walking” or “running” based on gait mode (i.e., absence or presence of flight phase). Receiver operator characteristic (ROC) curve analysis and confusion matrices were used to compute classification accuracy metrics (sensitivity, specificity, and accuracy) for the previously identified cadence thresholds in the current dataset. Youden’s Index (sensitivity + specificity - 1) was used to identify the optimal cadence for WRT thresholds in this sample. RESULTS: Cadence thresholds of 135 and 140 steps/min shared equal sensitivity (100%). However, specificity and accuracy were higher for the cadence threshold of 140 steps/min (95.5% and 96.6%, respectively) in comparison to 135 steps/min (92.4% and 94.3.6% respectively). Additionally, an optimal cadence for WRT prediction was 139 steps/min, yielding identical sensitivity, specificity, and accuracy (100, 95.5, and 96.6%, respectively) as the 140 steps/min threshold. CONCLUSIONS: These findings support the validity of 135 and 140 steps/min thresholds for predicting WRT, with the 140 steps/min threshold performing better overall in the current dataset. Future studies could explore how run-to-walk thresholds correspond with these cadence-based WRT thresholds.