case://study-002
AI / Full-StackProject 002
HealthFlow AI
HealthFlow AI is a hospital intelligence platform that combines admissions forecasting, precomputed daily predictions, and AI-generated operational guidance for smarter planning.
7 Days
Forecast
DB-first
Serving
Full Stack
Stack
AI / Full-Stack
Category
The Problem
What Wasn't Working
Hospital teams need early visibility into admissions trends, but real-time model inference inside request handlers can increase latency and make prediction delivery harder to scale reliably.
The Solution
How I Fixed It
I built a pipeline around historical inflow plus external signals, generated forecasts ahead of time, stored them in PostgreSQL, and served them through a Next.js + Flask architecture with AI advisory responses for explanations and staffing guidance.
Stack
Technologies Used
LightGBM
Next.js 14
PostgreSQL
Flask
Groq LLM
Vercel
Results
Key Outcomes
✓
7-day admissions forecasting workflow✓
Precomputed predictions instead of runtime inference✓
Database-first serving for lower request overhead✓
AI-generated trend and staffing recommendations