AMR Global Leadership Award (Grand Prize)
Team 13282: Anthonio Oladimeji Gabriel (Team lead, Centre for Clinical Intelligence and Safety) and Olowu Babatunde Ibrahim (Washington State University)
Global Burden–Funding Alignment and Optimization (GFAO): An Integrated Computational Pipeline for Resource-Allocation Optimization Against Multidrug-Resistant ESKAPE Pathogens
Billions in AMR R&D are deployed annually, yet investment and real-world clinical burden are imperfectly aligned. Rather than asserting an unidentifiable causal effect of funding on resistance, precluded by temporal mismatch, reverse causality, and the global, product-directed nature of R&D; GFAO quantifies the misalignment between where resistance burden concentrates and where research investment flows. We integrate four Vivli surveillance datasets (Pfizer ATLAS, Shionogi SIDERO-WT, Paratek KEYSTONE, SPIDAAR) with the Global AMR R&D Hub funding database through a four-stage pipeline.
First, because sponsor-run surveillance comprises convenience samples, we apply Deville–Särndal calibration weighting raked to WHO GLASS reference margins, with a pre-registered hierarchical fallback for sparse cells, reconstructing population-representative phenotypic baselines under EUCAST/CLSI breakpoints. Second, Latent Class Analysis groups isolates into stable multidrug-resistance phenotypes, tracking complex MDR such as XDR-CRAB as single epidemiological metrics.
Third, a Hierarchical Bayesian Multivariate Probit model, scoped to Klebsiella pneumoniae and Acinetobacter baumannii against three last-line classes (cefiderocol, omadacycline, carbapenems), projects resistance through 2030, borrowing strength toward data-sparse LMIC settings. Fourth, we construct a Global Misalignment Index comparing region-specific, severity-weighted clinical burden (SPIDAAR for LMICs; OECD references for high-income regions) against pathogen–drug-class funding shares, with pre-registered handling of unattributable platform funding under base and sensitivity scenarios.
Critically, each cell is cross-flagged against active clinical-pipeline candidates, distinguishing genuine neglect from rational, pipeline-served deprioritization. Outputs deploy via VALOR, an open-access Streamlit dashboard that transparently demarcates the modeled quantitative core from an illustrative allocation-counterfactual layer, letting funders and ministries simulate reallocation scenarios. All code is released openly. GFAO converts disconnected surveillance and funding data into an actionable, honest instrument for global resource optimization.