Status: Funded - Open
BACKGROUND: An estimated 1.3 – 4 million cases of cholera occur annually, and untreated severe infections are often fatal, especially for children. Accurate diagnoses is key to appropriate treatment, but gold standard diagnostics (PCR, culture) are generally unavailable in settings where cholera occurs. GAP: Rapid diagnostic tests (RDTs) exist for cholera, but their standalone performance is highly variable, making it unclear when it would be beneficial to use an RDT, and how to interpret the results when it is used. There is a need for a clinical prediction algorithm to triage the use and interpretation of cholera RDTs. HYPOTHESIS: We hypothesize that incorporating readily accessible epidemiologic and clinical data into a pediatric diarrhea decision support tool on when to use and how to interpret a cholera RDT will improve diagnostic performance compared to the current standard of practice. METHODS: We will analyze already-collected data (symptoms, demographics, lab results) from over 6,000 patients (half of which are children) from three study populations in Bangladesh from 2015, 2018, and 2023. We will conduct a secondary data analysis of these data and use machine learning to build predictive models. RESULTS: Pending. IMPACT: Positive results from this study will be validated in other settings and then incorporated into our existing pediatric diarrhea electronic decision support (eCDS) tool that automates guidelines for diarrheal disease management among children in low- and middle-income countries. Ultimately, we hope this will reduce pediatric diarrheal disease morbidity and mortality.