Hip fractures are among the leading cause of orthopedic hospitalization in the elderly population worldwide, resulting in significant health and financial burdens. Computer tomography (CT)-based finite element analysis (FEA) has emerged as a powerful tool for predicting the biomechanical response of femurs to assess hip fracture risk. However, FEA requires a qualified analyst to generate the necessary input data, verify the output and post-process the results for a meaningful conclusion. The aim of this work was to develop and validate a fully autonomous method that can be applied on clinical CT for hip fracture risk assessment. To address this, we developed Simfini [1], an autonomous CT-to-FEA software application for femurs that uses a deep learning (DL) algorithm to segment the femur of interest from CT scans, generate finite element mesh, and simulates two common loads that cause neck and intertrochanteric hip fractures. The FEA data are then processed with a machine learning (ML) algorithm that considers patients' weight, height, gender, and the biomechanical results at different regions along the femur. We conducted a retrospective clinical study to assess the performance of our approach in predicting hip fracture risk in patients with and without type 2 diabetes. This study used 418 clinical CT scans, with 239 without a fracture within five years after CT and 179 with a fracture. The coupled FEA and ML algorithm approach led to high accuracy in predicting the risk of hip fracture in both T2DM (sensitivity 92%, specificity 88%) and non-T2DM (sensitivity 83%, specificity 84%) populations. The result demonstrate that Simfini is a promising and innovative tool for autonomous hip fracture risk assessment using CT scans.