Foot orthotics are commonly prescribed to address lower limb conditions, yet understanding their biomechanical effects remains elusive due to conflicting research findings. Traditional statistical methods may overlook subtle changes induced by orthotics, prompting the need for innovative approaches. This study, involving 20 participants prescribed custom foot orthotics, leveraged machine learning alongside conventional statistics to analyze biomechanical gait data. While both methods identified key variables such as ankle and knee kinematics and kinetics, machine learning models consistently outperformed traditional statistics. Moreover, the interpretability of features in the summary machine learning pipeline makes it valuable for clinical applications. Despite limitations in sample size, this study underscores the potential of machine learning in enhancing the understanding and clinical application of foot orthotics. Continued research into machine learning and its potential applications for biomechanical data may offer valuable insights for optimizing orthotic design and prescription practice.