Probabilistic Inference of Comorbidities from Symptoms in Patients with Atrial Fibrillation: An Ontology-Driven Hybrid Clinical Decision Support System.

Atrial fibrillation $(AF)$ is the most prevalent cardiac arrhythmia. While $AF$ is a cardiological disease, its risk factors and mechanisms are often rooted in non-cardiological comorbidities, introducing complexity in the treatment of the heterogeneous patient population. This study presents the development of a clinical decision support system (CDSS), which aims to mitigate potential challenges of the cross-disciplinarity of $AF$ A knowledge base is implemented to capture the hierarchical nature of relevant concepts. $Nai\dot{v}e$ Bayes classifiers are used to predict the patient comorbidities related to $AF$ mechanisms and risk factors based on provided symptoms. The resulting CDSS infers comorbidities with a top-k accuracy of 0.53, 0.80, and 0.88 for $k=1, 3$ , and 5 respectively.