Machine Learning with Alpha Toxin Phenotype to Predict Clinical Outcome in Patients with Staphylococcus aureus Bloodstream Infection

Staphylococcus aureus bloodstream (SAB) infection remains a leading cause of sepsis-related mortality. Yet, current treatment does not account for variable virulence traits that mediate host dysregulated immune response, such as SA α-toxin (Hla)-mediated thrombocytopenia. Here, we applied machine learning (ML) to bacterial growth images combined with platelet count data predict patient outcomes. We profiled Hla phenotypes isolates collected from patients bacteremia by taking smartphone beta-hemolytic on sheep blood agar (SBA). Electronic medical records were reviewed extract relevant laboratory and clinical data. A convolutional neural network was process the plate image input along day 1 generate ML-based models thrombocytopenia 4 total 229 infected strains exhibiting varying zone sizes beta-hemolysis SBA included. 539 generated inputs model development. One-third developed at onset, an overall mortality rate 18.8%. The ML algorithm showed strong performance (AUC 0.92) predicting modest 0.711) Our findings support further development validation proof-of-concept application in digital microbiology, measure factor production carries prognostic significance can help guide selection.