This capstone project investigates how physiological sensor data from wearable devices can be used to model fatigue accumulation and predict cognitive load over time. Using two major datasets; Wearable Exam Stress and CLAS (Cognitive Load, Affect, and Stress). The study applies machine learning and deep learning techniques to extract meaningful features and forecast fatigue scores. Exploratory data analysis, clustering, and supervised learning methods (LSTM, Random Forest, XGBoost, SVM) were employed to capture trends and identify hidden fatigue patterns. The project demonstrates how sensor-based data science pipelines can support proactive fatigue management strategies in high-performance domains.