Prana Health

BSc (Hons) Computer Science | Software Development Group Project

Prana Health, is an enterprise-grade, AI-driven public health platform engineered to dismantle the reactive “”Time-Lag Trap”” of regional dengue management. While traditional health systems only deploy mitigation resources after hospitalizations spike, Prana Health provides actionable, localized outbreak forecasts up to four weeks in advance, shifting epidemiological intervention from crisis treatment to prevention. At its core, the platform utilizes an optimized XGBoost machine learning regression model. This algorithm ingests historical case counts, demographic data and localized meteorological parameters, such as rainfall and temperature lags, to calculate a dynamic “”Epidemic Ratio”” across 350+ Medical Officer of Health (MOH) divisions. XGBoost was explicitly chosen for its superior performance on structured tabular data and its “”white-box”” feature interpretability, giving public health inspectors clear mathematical proof of underlying biological triggers. Architecturally, Prana Health is built on a 100% serverless cloud ecosystem using Azure Functions and Azure SQL, maintaining near-zero operating overhead while scaling automatically during monsoon surges. The platform also features the “”Prana AI Analyst,”” a natural language interface powered by a custom LiteLLM routing engine that dynamically orchestrates top-tier LLMs (such as GPT-4o-mini and Llama 3.1) for zero-cost data interpretation. Prana Health operates a dual-sided model: a B2G visual dashboard empowers field inspectors to execute targeted larvicidal fogging, while the B2B “”Enterprise Adapter”” API enables private pharmaceutical supply chains to pre-position critical medical stock (like paracetamol and IV fluids) in impending high-risk zones weeks before local stockouts occur.