medbridgeai

byAkshay

Build a fully functional, premium healthcare AI web app called MEDBRIDGE AI for HE-05: Medical Document Intelligence & Patient Timeline. No login/signup. On first load, show a cinematic healthcare entry screen with an animated medical AI core, heartbeat waveform, floating medical documents and connected timeline nodes. Clicking the core opens the main landing page. Create an advanced professional health-tech UI using React, TypeScript, Tailwind, Framer Motion and Lucide icons. Use sophisticated navy/white/teal healthcare styling, smooth medical-data animations, responsive layouts, micro-interactions and meaningful transitions. Avoid gaming/neon aesthetics. Landing page hero: “From Fragmented Records to One Connected Patient Story.” Add [Explore Demo], [Upload Medical Records], [Load Demo Patient], [How It Works]. Include animated workflow: Documents → AI Extraction → Medical Events → Relationships → Timeline → Evidence. Build a real backend with FastAPI, PostgreSQL/SQLite fallback, OCR using PaddleOCR/Tesseract, OpenCV, PyMuPDF, structured AI extraction, FAISS/ChromaDB RAG and API-based translation/TTS with browser fallbacks. Never expose API keys. Implement working features: • Multi-PDF/JPG/PNG upload and processing • Document classification: lab, prescription, imaging, discharge, clinical note • OCR and typed + handwritten document processing • Medical information extraction: patient details, dates, medicines, dosage, lab values, procedures and documented clinical events • Confidence scores and human verification/editing • Cross-document event/relationship detection • Chronological interactive Patient Timeline with filtering/search • Patient 360: overview, medications, labs, procedures, documents and relationships • Evidence-first traceability: every extracted fact links to document/page/source and opens the original document • “Ask Your Records” RAG assistant answering ONLY from uploaded records, always showing sources; say “not found in records” when unavailable • Translation into English, Hindi, Telugu, Tamil, Kannada, Malayalam, Bengali, Marathi, Gujarati and Punjabi • Text-to-speech with animated waveform • Conflict detection showing conflicting values and both sources without deciding medically • Global search, notifications and professional analytics • Presentation Mode optimized for judges Create a synthetic demo patient “Arjun Rao” with 5 linked documents: laboratory report, prescription, imaging report, clinical note and discharge summary, including a handwritten-style document. Provide [START DEMO] that automatically demonstrates upload → processing → extraction → relationships → timeline → evidence → chatbot → Telugu translation → voice. Every button must work. No placeholder screens, fake statistics or dead interactions. If APIs are unavailable, implement a reliable DEMO MODE with deterministic synthetic data so the entire application still works. Add privacy and responsible-AI messaging: synthetic demo data only; the system organizes/retrieves documented information and does not diagnose or prescribe. Never hallucinate missing information. Preserve original records, confidence and uncertainty. Prioritize: P0: upload, document understanding, extraction, structured data, event ordering, relationships, timeline and source traceability. P1: handwriting, RAG assistant, multilingual, voice, confidence and verification. P2: conflicts, Patient 360, analytics, event graph, presentation mode and advanced animations. Create clean frontend/backend architecture, database models, API endpoints, README, .env.example and setup instructions. Test the complete application end-to-end, fix frontend/backend/API errors, console errors, broken routes, upload failures and responsiveness issues before finishing. The final result must look and function like a professional healthcare AI startup prototype, not a basic college CRUD project.