DIC03-TriPhar (TREE)

TREE is an AI-backed decision-support dashboard for pharmaceutical development. It helps discovery, clinical, and operations teams explore literature, clinical trials, pipeline data, regulatory alerts, and document analysis through a single interface—with live external APIs and optional LLM summarization.

---

What It Does

• Statistics — Search and explore clinical-trial and literature data with filters and visualisations.
• Insights (Decision Support Hub) — Pipeline monitoring (ClinicalTrials.gov–style), clinical document upload and AI summarisation (DeepSeek), side-by-side comparison of PDFs, and regulatory alerts / compliance-style reports.
• Trends — Predictions and trend views over merged medical/legal data.
• Research AI — Natural-language chat over TREE (query, role-aware summaries, citations). Optional streaming; chat history persisted per session.

External data is live API–only (no bulk download or storage of PubMed/ClinicalTrials.gov etc.); optional persistence is minimal (e.g. chat history, merged DB for analytics). See docs/DATA-HANDLING.md and docs/TREE-ARCHITECTURE.md for constraints and design.

---

Workflow

1. User — Logs in, picks a section (Statistics, Insights, Trends, Research AI) from the nav.
2. Frontend — Next.js app (React). All API calls go to /api/*, which Next rewrites to the backend.
3. Proxy — Next.js rewrites /api/:path* → http://127.0.0.1:8000/api/:path* (long requests use an increased proxy timeout).
4. Backend — FastAPI on port 8000: auth/CORS, routing, validation. Uses adapters for PubMed, ClinicalTrials.gov, merged DB, regulatory sources; DeepSeek for document summarisation, comparison, chat, and compliance summaries.
5. Data — Live API calls per request; optional SQLite DBs under data/ (e.g. merged medical data, legal/regulatory) for analytics and insights. No storage of full external API payloads.
6. Response — JSON (or stream) back to the frontend; UI renders tables, charts, markdown, and chat.

---

Repository Structure

DIC03-TriPhar/
├── app/                    # Next.js app router
│   └── page.tsx            # Root: nav state, login, and view switcher (home, statistics, insights, trends, research-ai)
├── components/
│   ├── global-navigation.tsx
│   └── views/              # Main screens
│       ├── login-view.tsx
│       ├── brand-gateway.tsx   # Home / landing
│       ├── statistics-view.tsx
│       ├── insights-view.tsx   # Pipeline, PDF analyse/compare, regulatory
│       ├── trends-view.tsx
│       └── research-ai.tsx     # Chat UI
├── backend/                 # FastAPI application
│   ├── app/
│   │   ├── main.py         # App entry, CORS, exception handler, router includes
│   │   └── config.py       # Settings (env, API keys, paths)
│   ├── api/
│   │   ├── routes/         # tree, insights, analytics, trends_ai, chat
│   │   └── schemas/        # Pydantic request/response models
│   ├── adapters/           # External data & LLM
│   │   ├── pubmed.py, clinical_trials_gov.py
│   │   ├── merged_db.py, regulatory_alerts*.py
│   │   ├── document_compare_llm.py, pdf_utils.py
│   │   └── chat_history.py
│   ├── agents/             # Literature, clinical, signals
│   ├── orchestrator/
│   └── models/
├── data/                   # Optional SQLite DBs (merged medical, legal); not committed
├── docs/                   # Architecture, data-handling, status
├── next.config.mjs        # Rewrites to backend; proxyTimeout for long /api calls
├── package.json            # Next.js, React, UI deps
└── README.md              # This file

---

Quick Start

1. Backend (from repo root):

   cd backend
   python -m venv .venv
   .venv\Scripts\activate   # Windows
   pip install -r requirements.txt
   cd ..
   uvicorn backend.app.main:app --reload --host 0.0.0.0 --port 8000

   • API: http://localhost:8000
   • Docs: http://localhost:8000/docs

2. Frontend:

   npm install
   npm run dev

   • App: http://localhost:3000 (or the port Next prints).
   • Set DEEPSEEK_API_KEY in backend/.env for PDF summarisation, chat, and compliance LLM features.

3. Optional — Ensure data/merged_medical_data.db (and any other DBs your flows need) exist if you use pipeline/trends features that depend on them.

For more detail (endpoints, env, data rules), see backend/README.md and git**docs/**.