Clinic-Buddy

• 

  Process Flow Diagram

Inspiration:

90% of clinical trials fail, costing billions and delaying life-saving treatments. We wanted to give researchers a tool to predict outcomes before investing massive resources—turning guesswork into data-driven decisions.

What it does:

Clinic-Buddy is an AI-powered clinical trial simulator. Users input trial parameters (drug, phase, patients, therapeutic area), and four specialized AI agents analyze the data to predict success probability, identify risks, and provide actionable recommendations. It also features a multilingual AI chat assistant and image analysis for trial setups.

How we built it:

• Backend: Python FastAPI with AWS Bedrock (Claude Sonnet 4.5)

• AI Agents: Patient, Drug, Trial, and Regulatory agents working in parallel

• Frontend: Tailwind CSS, Alpine.js, Chart.js for visualizations

• Database: CSV-based storage for rapid prototyping

• Deployment: Docker containerized, deployed on Google Cloud Run

Challenges we ran into:

• Handling AWS Bedrock API timeouts for complex image analysis (solved with 300s timeout config)

• Coordinating four AI agents to produce coherent, unified predictions

• Building a responsive multilingual chat that auto-detects user language

• Persisting analysis data reliably across sessions

Accomplishments that we're proud of:

• Four AI agents working together to simulate real clinical trial evaluation

• Support for 10+ languages in chat assistant

• Image analysis feature for trial setup evaluation

• Clean, professional dark-themed UI

• End-to-end working prototype in limited time

What we learned:

• How to orchestrate multiple AI agents for complex decision-making

• AWS Bedrock integration patterns and timeout handling

• Importance of graceful fallbacks when AI services are unavailable

• Building intuitive UX for complex scientific workflows

What's next for Clinic-Buddy:

• Integration with real clinical trial databases (ClinicalTrials.gov)

• Machine learning models trained on historical trial outcomes

• PDF report generation for stakeholder presentations

• Real-time collaboration features for research teams

• Mobile app for on-the-go trial monitoring

Built With

• alpine.js
• amazon-web-services
• chart.js
• claude
• docker
• fastapi
• multi-agent
• python
• sonnet-4.5
• tailwind