Research Papers

Dan Griffiths

This paper presents a field-deployable framework that converts synchronized IPhone video into predictive neuromuscular performance insights for track and field athletes. Using markerless motion capture and musculoskeletal optimization, the study reconstructs internal joint torques, muscle activations, and ground-reaction forces from video alone. These neuromechanical features are integrated with machine learning models, (XGBoost. RNN, LSTM-NN)  to predict javelin throw distance with high accuracy (R² = 0.92). The results demonstrate that scalable, video-based biomechanics can move sports analytics from descriptive observation to predictive, athlete-specific intelligence.