Antimicrobial resistance (AMR) is one of the most urgent threats to global health today. When antibiotics lose effectiveness, common infections become harder to treat, resulting in longer illnesses, increased healthcare costs, and a higher risk of death. To tackle this issue, we must understand how resistance develops and spreads across communities and healthcare systems.
Our proposed project will utilize two large global datasets from the Vivli AMR Register: the GSK SOAR 201910 and Pfizer ATLAS datasets. Combining them, we can create a more comprehensive picture of resistance trends across regions, age groups (including children), and over time.
This project will:
- Improve patient outcomes by identifying the effective antibiotics in various regions, enabling better treatment decisions.
- Strengthen antimicrobial stewardship by modeling how common prescribing practices contribute to resistance, assisting in guiding responsible antibiotic use.
- Inform public health practices by identifying areas where resistance is increasing and highlighting emerging threats that require immediate attention.
- Support health systems by providing evidence to shape policies, update treatment guidelines, and inform antibiotic procurement based on real-world data.
Using advanced machine learning techniques, we will develop a predictive model of AMR to forecast susceptibility trends across regions, age groups, and antibiotic classes. We will explore how clinical and community-level factors, such as prescribing practices, population characteristics, and health system interventions, influence the emergence and spread of resistance. Importantly, we will generate interpretable and actionable outputs, including resistance risk scores and antibiotic effectiveness predictors, and present these insights through a user-friendly interactive dashboard to support decision-making by healthcare workers, policymakers, and public health officials.
By transforming surveillance data into predictive tools, we aim to promote prudent antibiotic use, strengthen AMR preparedness, and protect communities. The modeling framework and tools developed will be adaptable for use by national and regional health systems, enabling scalability and long-term impact.