AMR Triage

Background With high burden of antimicrobial resistance and steep increase in watch antibiotics in low and low-middle income countries (LMICs), a major hurdle in antibiotic stewardship measures is limited diagnostic capabilities. We aim to develop an AI tool to calculate a patient’s probability of having or developing antibiotic resistance based on features such as the geography, patient characteristics and infection type features, with focus on any bacterial infection (excluding tuberculosis). Methods A prediction model would be built in the form of a webtool: 1. Variables include: Country, Specimen/Isolate date, Age, Patient location, Infection type, Causative pathogen, acquisition type, Antibiotic susceptibility results, genotyping results among others. 2. Data sources: Data from LMICs from the Pfizer’s ATLAS program and Venatorx’s GEARS 3. Prediction model: Using the surveillance data mentioned above, the best performing model would be selected to predict likelihood of resistances. Missing priors would be obtained from other publicly available surveillance/published results from well-powered studies. Expected Results A web based triage tool based on analysis results, providing country or continent specific (depending on the data coverage) prediction of antibiotic resistance and list of safe antibiotics e.g. a list of highest probable pathogens and antibiotic resistances and, a list of recommended antibiotics for a 60-year old patient suspected to have contracted a nosocomial urinary tract infection in a medical ward in East Africa. Validation and performance evaluation would be planned as a next phase using available hospital level data in LMICs. Expected use of the results The “AMR-triage” will be a readily available user-friendly tool for clinicians in any outpatient or inpatient setting. It will optimize the use of limited resources for laboratory testing, prioritizing patients to be tested. The tool will guide the empirical treatment and red flag any inappropriate antibiotic prescription, thus supporting antibiotic stewardship efforts.