intelligent-healthcare-data-ai-platform

The Intelligent Healthcare Data & AI Platform is an end-to-end system for analyzing patient encounters, predicting 30-day hospital readmission risk, and generating grounded healthcare analytics insights. It uses fully synthetic healthcare data and is intended for engineering and portfolio demonstration purposes only, not as a clinical decision-support system. The platform addresses the business problem that hospital readmissions can increase healthcare costs and may indicate gaps in discharge planning and follow-up care. It demonstrates how healthcare encounter data can be transformed into analytics-ready datasets, used to train a machine learning risk model, and connected to a Retrieval-Augmented Generation (RAG) assistant for grounded analytical Q&A. The data engineering pipeline follows a Bronze, Silver, and Gold medallion architecture, with a separate PySpark and Delta Lake implementation for Databricks. A Random Forest classifier predicts 30-day hospital readmission risk using patient encounter and clinical features, and the trained model is persisted with Joblib and served through a FastAPI REST service. The project also implements a fully local Retrieval-Augmented Generation pipeline using Ollama, which retrieves relevant healthcare analytics records before sending context to the LLM, reducing unsupported responses and grounding answers in the processed dataset.