Bayesian Neural Network to Predict Antibiotic Resistance

International audience Antimicrobial resistance is recognized by the World Health Organization (WHO) as a significant global health threat. The accurate identification of bacterial susceptibility to antibiotics is crucial, but it often takes several days. On the other hand, in medical decision support systems, such as the one proposed in this contribution, it is crucial to assess the uncertainty of the model when a decision is provided. In this work, we propose a model based on a Bayesian Neural Network to predict antibiotic resistance at different stages of the antibiogram process for a set of 47 antibiotic therapies. Excellent results were achieved, with the area under the receiver operating curve reaching up to 0.9 at the final stage, while also providing a measure of the epistemic uncertainty. To enable clinical usage of the proposed approach as a decision support system, the model has been integrated into a user-friendly and responsive web application accessible on both mobile phones and desktops.