Developed a novel SEIRS-based Physics-Informed Neural Network (SEIRS-PINN) for climate-driven malaria forecasting by integrating epidemiological dynamics with deep learning. Designed a two-stage architecture incorporating climate variables, seasonality, and transmission dynamics to learn biologically consistent disease progression while preserving SEIRS compartment constraints. The framework achieved a test R² of 0.71–0.75 with approximately 20% MAPE, while providing uncertainty quantification through Monte Carlo Dropout and maintaining full biological validity. Conducted extensive benchmarking against classical machine learning and epidemiological models, demonstrating an effective trade-off between predictive accuracy, interpretability, and epidemiological realism. 

Worked under Dr. Purnedu Mishra & Postdoctoral Researcher Dr. Himanshu Jain