MedSim-AI

An AI-driven medical simulation website that trains healthcare professionals through interactive patient encounters, diagnostic evaluation, symptom analysis, and decision-tree exercises.

Table of Contents

1. Overview
2. Features
3. How It Works
4. How To Run
5. User Guide

Overview

MedSim-AI simulates realistic clinical scenarios using Large Language Models (LLMs) and machine learning. Learners can:

• Conduct patient interviews in natural language
• Receive AI-generated symptom profiles and histories
• Submit diagnoses and receive structured feedback
• Navigate interactive decision trees for differential diagnosis

Features

• Virtual Patient Chat

  • LLM-powered responses with configurable patient profiles

• Diagnostic Evaluator

  • Scoring on medical knowledge, communication, presentation

• Synthetic Symptom Generator

  • Randomized or disease-specific symptom sets

• Decision-Tree Module

  • Branching logic guided by user inputs

• Semantic Symptom Matching

  • Symptom–disease similarity for prediction

How It Works

1. Select Scenario
   • Choose from preset diseases or generate random profiles
2. Interview Patient
   • Ask questions and receive detailed responses
3. Submit Diagnosis
   • Enter differential or final diagnosis
4. Evaluate & Learn
   • View graded feedback and review decision paths

How To Run

Frontend

git clone https://github.com/Abhigyan126/MedSim-AI.git
cd MedSim-AI
cd src/front/medsim-ai-front/
npm start

Backend

Create .env in src/back/ and fill in the following credentials

mongo  = 'API key for mongodb'
secret = 'Secret key for JWR'
key = 'API key for gemini'

cd MedSim-AI
pip install -r requirements.txt
cd src/back
python main.py

Setup Virtual Environment

python -m venv 'name of environment'
# for Linux or mac os
source 'name of environment'/bin/activate
# for windows os
'name of environment'\Scripts\activate

User Guide

Virtual Patient Simulator

• Generate random or specific diseases with customizable patient profiles
• View symptoms on human anatomy diagram
• Ask questions through chat interface
• Submit and receive feedback on diagnoses

Diagnostic Evaluator

• Submit diagnosis and rationale
• Receive feedback on medical competency, communication, and presentation

Decision-Tree Explorer

• Answer yes/no questions for guided diagnosis
• Navigate decision paths with backtracking capability

DiseaseCraft

• Match symptoms to target diseases
• Drag and drop interface with instant scoring feedback