"Data Challenge : ProxySteward - A zero-cost framework for predicting CRKP and Prioritising AMS in Kenya."

Proposed Research Question and Objectives Can cross-referencing industry surveillance data with Global R&D funding patterns identify Carbapenem Resistant Klebsiella Pneumonia (CRKP) hotspots to dynamically prioritize local antimicrobial stewardship (AMS) in Kenya? This project will build a scalable, python-based predictive framework that models CRKP prevalence and directly ranks healthcare facilities, ensuring limited AMS resources are deployed where clinical risk is highest and funding is lowest. Datasets to be used We will utilize the Pfizer ATLAS dataset to extract longitudinal MIC trends, specimen categories and resistant profiles for K. pneumonia in Kenya. To capture critical gaps, we will cross-reference this with the Global AMR R&D Hub repository to analyze regional stewardship funding, national action plan, investments and resource pipeline distribution. Methodology and Analytical Approach Using python (pandas, scikit-learn, geopandas) we will clean and filter Kenyan data inputs. Instead of just mapping resistance, we will build a stewardship prioritization algorithm. This pipeline weights CRKP probability against localized R&D deficits. The framework will generate a dynamic ”AMS priority index score” for different areas, automatically flagging regions with surging CKRP but logging stewardship infrastructure. Impact and team capabilities The framework translates data directly into an actionable development roadmap for public health decision-makers. It answers not just where the bacteria is but exactly where to send infection control teams tomorrow. Its designed for scalability across Sub –Saharan Africa. Our multidisciplinary team guarantees success. It consist of ICT specialist for pipeline architecture, a Statistician for prioritization algorithm, a Researcher for data integration and a frontline Nurse ensuring the priority index aligns with bedside clinical workflow and hospital resource constraints