Finite element analysis (FEA), introduced more than half a century ago, requires a (qualified) analyst to generate the necessary input data, verify the output and post process the analysis results for a meaningful conclusion. The required expertise and labor efforts have precluded the use of FEA in daily medical practice by orthopedic surgeons for example. Patient-specific analyses of the mechanical response of human bones may have a tremendous impact in clinical practice should they be easily accessible by orthopedic surgeons. Recent scientific advancements such as low dose CT scans, machine learning, and high order FEA which facilitates an inherent methodology for assessing numerical accuracy allow a fully autonomous process for assessing bone strength and fracture risk. This autonomous process, that we shall refer to here as “Autonomous Finite Element” (AFE) analysis, introduces a paradigm shift in the use of FEA. We shall describe herein a novel process that utilizes AFE to produce a patient-specific assessment of bone strength. The process consists of an automatic segmentation of femurs from CT-scans by convolution neural networks, an automatic mesh generation and application of boundary conditions based on anatomical points, a high-order FE analysis with numerical error control, and finally an automatic report with a clear assessment of bone fracture risk. One specific application of AFE is the determination of the risk of fracture for patients with tumors of the femur and whether a prophylactic surgery is needed. To the best of our knowledge this is the first CE accredited AFE application being used by orthopedic surgeons in clinical practice.