<p>To enhance the accuracy and objectivity of rehabilitation assessment for lower-limb injuries which are often hampered by inter-rater variability and subjectivity, this research work leverages both Artificial intelligence (AI) and optimization techniques. By employing a novel fusion-based metaheuristic optimization strategy (FGAPSO), this study identifies optimal gait features critical for predicting rehabilitation duration. Using an open-access dataset of 2084 patients with functional disorders (including ankle, knee, hip, and calcaneus injuries), the selected features are integrated into ensemble model (EM) which are further optimized using simplex method. In addition, exponential function is used to study improvement or recovery trajectory by utilizing composite gait metric developed from selected optimal features. Hence, the proposed methodology significantly improved prediction accuracy using baseline gait trajectories. Comparative analysis of heeled gait trajectories with baseline and normative data of healthy controls underscores the framework’s effectiveness with an accuracy of 94.7 ± 1.2%. This research directly contributes to Sustainable Development Goal (SDG), good health and well-being by supporting timely rehabilitation through identified relevant gait features and prediction of optimal post-surgery rehabilitation duration aiding the return to normal walking. Thus, the carried work presented a cost-effective objective solution for clinical gait analysis reducing subjectivity. It provides clinicians with an accurate, automated, and reliable tool for predicting effective rehabilitation duration ultimately resulting in improved patient outcomes and planning of more efficient rehabilitation procedures.</p>