• 

Inspiration

Healthcare accessibility remains a major challenge, especially in rural and underserved areas. Many people struggle to understand their symptoms or medical reports without professional help. This inspired us to build an intelligent system that can assist users in making informed health decisions quickly.

We wanted to create something beyond a simple app — a reusable AI capability (superpower) that any AI agent can use. This led us to adopt the MCP (Model Context Protocol) approach, enabling modular and scalable healthcare intelligence.

What We Learned

During this project, we learned:

How to design and build an MCP server architecture

Creating modular AI tools that can be reused by different agents

Handling healthcare data and converting it into human-readable insights

Designing APIs for real-world applications

Improving user experience by simplifying complex medical information

How We Built It

We developed a backend MCP server that exposes three core AI tools:

1. Symptom Checker

Analyzes user symptoms and predicts possible diseases, risk levels, and suggests the appropriate doctor.

1. Health Risk Predictor

Estimates risks such as diabetes and heart disease using parameters like age, BMI, and glucose levels.

1. Smart Medical Report Analyzer

Processes medical report data and identifies abnormal values while providing simple explanations.

Architecture User → Prompt Opinion Agent → MCP Server → AI Tools → Response Example Logic (LaTeX)

The system uses simple rule-based conditions such as:

𝑅 𝑖 𝑠

𝑘

{ 𝐻 𝑖 𝑔 ℎ

if glucose > 140

𝐿 𝑜 𝑤

otherwise Risk={ High Low ​

if glucose > 140 otherwise ​Challenges We Faced

Understanding and implementing the MCP architecture

Designing accurate yet simple health prediction logic

Converting complex medical data into easy-to-understand insights

Ensuring modularity so tools can be reused independently

Handling edge cases and invalid user inputs

Key Features

Modular AI tools (plug-and-play via MCP)

Real-time symptom analysis

Health risk prediction

Intelligent medical report interpretation

Scalable and reusable architecture

Future Scope

Integration with real medical datasets and ML models

PDF report upload and automatic extraction

Multilingual support (English + Telugu)

Voice-based interaction

Personalized health tracking

Built With

• express.js
• javascript-(node.js)
• json
• mcp
• mongodb-(optional)
• openai/gemini-apis
• pdf-parse
• postman
• prompt-opinion
• render/vercel/aws
• vs-code