Adaptive Footwear Stiffness Driven by Biomechanical Signals for Improving High-Intensity Metabolic-Neuromuscular Performance: A Randomized Controlled Trial

Background: Basketball shooting performance deteriorates under fatigue, often due to compromised biomechanics and suboptimal footwear stiffness. Footwear with dynamically adjustable longitudinal bending stiffness (LBS) could counteract fatigue effects by maintaining optimal support throughout play. Objective: The objective of this study is to evaluate whether an adaptive LBS basketball shoe—controlled by a machine learning algorithm (logistic regression with support vector machine–recursive feature elimination) using biomechanical signals—can improve high-intensity shooting performance under fatigue compared to the fixed-stiffness footwear. Methods: A total of 60 participants were randomly assigned to high-stiffness (HS), low-stiffness (LS), or self-adaptive stiffness (SS) shoe groups. All completed a 2-min high-intensity shooting and rebounding drill designed to induce fatigue, while real-time kinematic data of the lower limbs and physiological data (heart rate, muscle oxygen saturation) guided SS stiffness adjustments. Results: The SS group achieved significantly more successful shots than the HS and LS groups (mean difference +1.662 vs HS, p < 0.01; +2.753 vs LS, p < 0.001) and higher shooting accuracy (+0.093 vs HS, p < 0.01; +0.117 vs LS, p < 0.001). Under fatigue, SS footwear preserved favorable lower-limb joint kinematics (e.g., maximum hip rotation angle +0.234 rad vs HS, p = 0.014) without increasing cardiovascular or metabolic demands (no significant differences in heart rate or SmO2 ). Conclusion: Adaptive footwear stiffness integrating biomechanical sensing and machine learning improved basketball shooting performance and mitigated fatigue-induced biomechanical degradation, highlighting its potential for enhancing sports performance. Received: 25 November 2025 | Revised: 26 January 2026 | Accepted: 25 February 2026 Conflicts of Interest The authors declare that they have no conflicts of interest to this work. Data Availability Statement The data that support the findings of this study are openly available in the supplementary files of this article. Author Contribution Statement Yining Xu: Conceptualization, Methodology, Formal analysis, Investigation, Resources, Data curation, Writing – original draft, Visualization. Yang Song: Conceptualization, Methodology, Formal analysis, Investigation, Resources, Data curation, Writing – original draft, Visualization. Dong Sun: Conceptualization, Validation, Writing – review & editing, Supervision, Project administration, Funding acquisition. Zhiyi Zheng: Conceptualization, Methodology, Formal analysis, Investigation, Resources, Data curation, Writing – original draft, Visualization. Mingwei Sun: Conceptualization, Methodology, Formal analysis, Investigation, Resources, Data curation, Writing – original draft, Visualization. Wenlong Li: Methodology, Software, Writing – review & editing. Jiachao Cai: Methodology, Software, Writing – review & editing. Xuanzhen Cen: Methodology, Software, Writing – review & editing. Zixiang Gao: Methodology, Software, Writing – review & editing. Liangliang Xiang: Methodology, Software, Writing – review & editing. Monèm Jemni: Methodology, Software, Writing – review & editing. Yaodong Gu: Conceptualization, Validation, Writing – review & editing, Supervision, Project administration, Funding acquisition.