Atypical Femur Fracture diagnosis and Risk Assessment using Artificial Intelligence and global Data pooling for optimized fragility fracture prevention - the AFRAID study

Skeletal fragility is highly prevalent in the older population, with 1 in 2 women expected to sustain a fragility fracture in their lifetime - a risk that is substantially reduced with drugs preventing bone loss (bisphosphonates - BP).

Following reports of atypical femoral fractures (AFFs), the utilization of these effective drugs has decreased by 50% since 2005. An AFF is a rare, unfortunately often-missed type of stress fracture of the thigh bone. If treated inadequately, AFFs can cause serious complications.

In this 5-year study, we aim to:(1) improve the diagnostic accuracy of AFF using artificial intelligence (AI);(2) assess the risk for AFF compared to the risk reduction for fragility fracture for different strategies of BP treatment.

In Years 1 and 2, we will pool our existing fracture cohort with those of our renowned international collaborators, while extending the Swedish cohort using a novel nation-wide image extraction method.

In Years 3–5, we will apply a `fusion approach`, combining register and imaging data, to create an AI application helping physicians to identify AFF.

In parallel, we will use Cox models to estimate the risk versus benefit of BP treatment by discriminating the effects of treatment duration, drug holidays, and individual risk factors.

The outcomes will inform clinical decision-making through precision medicine while decreasing the frequencies of preventable thigh bone fractures and AFF and reduce complications associated with an AFF.