MediVerse AI

This project builds LLM-powered AI agents that work like a multidisciplinary medical review team. Each agent looks at the same medical report from a different clinical angle, then their findings are combined into a final summary of three likely health issues with reasoning.

This is meant to show how AI agents can collaborate across specialties and reason about complex medical cases.

⚠️ Note: This is strictly for research and learning.
It is not a medical device and must not be used for real diagnosis or treatment.

---

🚀 How It Works

Right now the system runs three specialist AI agents (GPT-5) in parallel using Python threading. Each agent reviews the report independently and returns insights related to its domain. The system then aggregates everything into a structured summary.

The AI Agents

1. Cardiologist Agent
Focus: Possible cardiac issues such as arrhythmias or structural problems.
Recommends testing, monitoring, and management strategies.

2. Psychologist Agent
Focus: Psychological contributors like anxiety or panic disorder.
Recommends therapy, stress-reduction or medication review where appropriate.

3. Pulmonologist Agent
Focus: Respiratory causes such as asthma or breathing disorders.
Recommends lung-function testing and respiratory treatment options.

---

📂 Repo Layout

• Medical Reports/ — synthetic sample reports
• Results/ — generated outputs from agents

---

⚡ Quickstart

1. Clone the repo
   git clone https://github.com/ahmadvh/MediVerse-AI.git
   cd MediVerse-AI

2. Create a virtual environment and install dependencies
   python -m venv venv
   source venv/bin/activate # Windows: venv\Scripts\activate
   pip install -r requirements.txt

3. Add your API key
   Create a file called apikey.env in the project root with:
   OPENAI_API_KEY=your_api_key_here

4. Run the system
   python main.py

---

🔮 Roadmap

Planned upgrades include:

• More specialist agents (Neurology, Endocrinology, Immunology, etc.)
• Local LLM support via Ollama / vLLM / llama.cpp
• Vision-based agents for radiology and medical imaging
• Structured dataset + live search tools
• JSON-structured outputs with validation
• Automated testing and reproducibility with mocked LLM calls

---

This project explores how AI agents can collaborate the way real clinical teams do — purely inside a research sandbox.