
TwinCare AI - Personal Health Intelligence Platform Every year, millions receive blood reports filled with numbers they can't interpret, hiding early warning signs of serious disease in plain sight. TwinCare AI changes that. It's an AI-powered health platform that turns raw lab reports into a living Digital Twin - a continuously evolving digital mirror of your health that grows smarter with every upload. Upload a blood report as a PDF or image, and TwinCare AI's two-stage pipeline extracts 35+ biomarkers using OCR and medical pattern matching, with an LLM fallback (Gemma 2) for complex formats. Each biomarker is validated against clinical ranges and classified as normal, low, high, or critical. The Digital Twin then recalculates personalized health scores across six organ systems - Heart, Liver, Kidney, Blood, Metabolic, and Thyroid - giving users a clear, at-a-glance wellness picture. But TwinCare AI goes beyond tracking. A calibrated LightGBM model predicts 10-year heart disease risk, and SHAP-powered explainability reveals exactly which biomarkers are driving risk up or down - no black boxes. A conversational AI Health Copilot, grounded in the user's own data, answers questions like "What does my cholesterol mean?" with personalized responses instead of generic advice. Built for production with FastAPI, PostgreSQL, React, and Docker Compose, the ML pipeline was trained on AMD Instinct MI210 GPUs using ROCm. The future potential is transformative. TwinCare AI is designed to evolve into a one-stop platform for all health needs: multi-modal ingestion of X-rays, CT scans, and wearable data; predictive models for diabetes, kidney, and liver disease; health trajectory simulations; medical literature retrieval; EHR interoperability via HL7 FHIR; and multi-language support. One platform where anyone can track, understand, and take control of their health - powered by explainable AI.
13 Jul 2026