Comparing K-Means Clustering and Youden’s J Statistic for Determining Y-Balance Test Cut-off Values for Classifying Chronic Ankle Instability in Logistics Workers with a History of Ankle Lateral Sprains

Background: Chronic ankle instability (CAI) is a common condition among logistics workers (LWs) that can significantly impact workplace productivity. Accurate classification of CAI using the Y-Balance Test (YBT) is crucial for effective management and timely return to work. Purpose: To compare the effectiveness of K-means clustering and Youden’s J statistic in determining YBT cut-off values for classifying CAI in LWs with a history of ankle sprains. Study design: Retrospective cohort study Methods: Data from 121 LWs with a history of ankle sprains were analyzed. YBT measures included anterior, posterolateral, and posteromedial reach distances, and composite scores. Cut-off values were determined using Youden’s J statistic and two K-means clustering approaches (Mean and Top 2). Performance metrics including area under the curve (AUC), sensitivity, specificity, and odds ratios were calculated for each method. Results: The YBT posteromedial direction distance, using Youden’s method, demonstrated the highest discriminative ability for CAI classification (AUC: 0.62, OR: 5.41, 95% CI: 2.09–14.04). The K-means Top 2 method consistently provided higher cut-offs with improved specificity across all YBT measures, notably achieving 87% specificity for the YBT composite score (cut-off: 96.98%). Conclusions: While both methods effectively identify CAI risk, the K-means clustering approach, particularly the Top 2 method, offers higher cut-offs with improved specificity. This suggests potential benefits in occupational health settings where stringent screening criteria are necessary for early identification and management of CAI risk in LWs.