Artificial intelligence support for more accurate diagnosis of lateral malleolar fractures

The incidence of lateral malleolar fractures (LMF) is increasing, and current classification systems have poor prognostic value in evaluating stability of these fractures[1] This often relegates surgical diagnoses using these systems to choose between operative and non-operative treatment for LMFs not more accurately than the toss of a coin. Better strategies and methods are needed in clinical decision making to include improvements in existing diagnostic procedures using Artificial Intelligence (Al) [2]. To specifically improve LMF diagnoses we introduce an Albased method which improves physician data interpretation of uncertainty [2], helping to reduce inaccurate inferences from patient data and improving physician predictive capability. In this paper we leverage existing, Al-based computer technology [3] and outline an improved, semi-automated solution. Several goals are developed: Identify a set of prognostic/predictive features to improve LMF classification, assign multiple degrees of uncertainty confidence supporting complex classification of LMF to aid improved physician diagnostic understanding. We believe our approach can lead to clinical validation, potentially incorporation of an improved algorithm into a mainstream LMF classification system, and application in broader clinical diagnostic settings. Since only an outline of the approach can be given in this short paper; sufficient details directed towards a more complete solution as follow-on will be identified leading to better assessment, diagnosis, and ultimately the more successful treatment of LMF.