Finalist
Team 13305 from Manipal University College Malaysia: Basavaraj Ramappa Poojar (Team lead) and Nagalakshmi Narasimha Swamy
Every clinician has watched it happen: a patient admitted with one infection leaves with another. A broad-spectrum antibiotic clears the original pathogen, but in doing so strips away the gut flora that kept resistant organisms in check. Clostridioides difficile, vancomycin-resistant Enterococcus, MRSA — these secondary infections are not coincidental. They are the ecological cost of the prescription, and right now, no tool exists to tell a prescriber how high that cost is before they write it.
Our research addresses this directly. We call it the ‘ecological AMR risk’ of an antibiotic — its propensity to cause secondary drug-resistant infections beyond killing its target. This risk is well-understood mechanistically but has never been measured at scale using real-world data. The result is a stewardship blind spot: formularies rank antibiotics by efficacy and toxicity, but not by the resistance they breed in the same patient.
We will mine the FDA’s Adverse Event Reporting System (FAERS) — over 20 million reports from 130 countries — for MedDRA-coded signals of secondary resistant infections following six antibiotic classes. Using pharmacovigilance disproportionality methods (Proportional Reporting Ratio and EB05 Bayesian scoring), we will quantify which antibiotics generate the strongest real-world signals for C. difficile, VRE, MRSA, and ESBL-associated events. These signals will then be cross-validated against minimum inhibitory concentration data from the Vivli AMR Register — ATLAS, SMART, KEYSTONE, GEARS, and SOAR — to establish them as surrogate markers for ecological AMR risk.
The entire analysis runs on open-source software: R with the PhViD package for signal statistics, and Python (pandas, SciPy, Seaborn) for data extraction, cross-validation, and visualisation.
The primary output is a ranked Antibiotic Ecological Risk Table — a practical, evidence-based score for each drug class that clinicians and stewardship teams can use at the point of prescribing. For patients, it means fewer secondary resistant infections. For stewardship programmes, it fills a measurable gap in how antibiotic choices are justified. For public health, it provides a scalable, laboratory-free surveillance signal that works even where MIC testing capacity is limited. For health systems, it offers a reusable, open-source framework — published on GitHub — that any institution can adapt without additional cost.