AI-DRIVEN BIOMECHANICAL OPTIMIZATION OF 100–200 M SPRINT PERFORMANCE A CONTROLLED EXPERIMENTAL STUDY

Аннотация

This study investigates whether AI-driven biomechanical optimization can enhance 100–200 m sprint performance in a controlled experimental setting. Sixteen trained sprinters were randomized into an Experimental group (n = 8) and a Control group (n = 8).

Over eight weeks, the Experimental group received real-time AI-informed feedback and individualized technique optimization guided by multimodal biomechanical data (kinematics, kinetics, and muscle activation proxies) collected via high-speed video, inertial measurement units (IMUs), and portable force sensors. AI models included regression-based predictors, sequence models, and Bayesian optimization to identify target changes in key gait and sprint mechanics (e.g., step frequency, step length, wall contact power, hip and knee extension angles).

The primary outcome was 100 m time (and 200 m as a secondary measure). Secondary outcomes included changes in velocity profiles, stride characteristics, joint kinematics, and ground reaction force metrics. Results indicated that the Experimental group achieved greater improvements in 100 m time compared with controls (mean difference = X.XX s, p < .05) and showed favorable shifts in velocity trajectories, step characteristics, and propulsive impulse. The AI-driven protocol demonstrated robust within-subject improvements across multiple biomechanical variables and provided practical guidelines for coaching application. Limitations include sample size, short intervention duration relative to season-long training, and the need for field validation across different surfaces and wind conditions. Overall, the study supports the potential of AI-assisted biomechanical optimization to augment sprint performance in a controlled experimental framework, with implications for precision coaching and individualized training planning.