Ankle injury prediction in motor vehicle crashes poses challenges due to the chaotic nature of ankle loading and the breadth of loading modes possible. The study seeks to refine the ankle joints in the THUMS v4.1 small female model and compare the biofidelity of the model against available Post-Mortem Human Subjects (PMHS) data. First, cartilage was added using hexahedral elements on the articulating surfaces to fill any voids between the bones and to improve continuity of load transfer. Next, a Mil-Spec shoe model was fitted to the foot by a systematic process. The refined and updated model was evaluated by performing four biofidelity validation cases - compressive axial impact, dorsiflexion, inversion, and eversion - replicating experiments conducted on small female PMHS with the same shoes. Also, a sensitivity analysis was performed on the material model of the ankle ligaments to understand the effect on the model response. The refined model showed consistency with the PMHS test results over the suite of validation cases performed. From the results, it was observed that the interaction between the bones has improved with the addition of cartilage resulting in softer response due to reduced rigid contacts. The addition of the shoe also enhanced the realistic simulation of ankle loading in likely real-world scenarios.