Hacktoberfest Build for a Friend: AI Symptom Checker for Timely Doctor Visits

Keshav Gilhotra — Sun, 04 Oct 2026 19:11:27 +0000

This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend

What I Built

We often find ourselves or our loved ones feeling unwell with a mix of confusing symptoms—like a sudden mild fever combined with joint pain, chills, or stomach discomfort—and we have no idea which disease those symptoms might be pointing to.

I built this project for my friends and family so that instead of panicking or blindly searching on Google, they have a quick, intelligent helper to analyze what might be going on. By identifying potential risks early, they can get timely clarity and consult a doctor to get the right prescription and treatment as soon as possible before things get serious.

⚠️ Important Note: This tool is strictly an educational helper and is never meant to replace professional doctors or medical experts. Always consult a certified healthcare professional for medical advice, and use this tool at your own discretion.

Demo

GitHub Repository: https://github.com/KrishTrue/dev-challange-1
Features in Action:
- Interactive symptom picker categorized across 130+ body symptoms
- Instant disease prediction with detailed explanations (symptoms, causes, precautions, and general medication guidance)
- Real-time chat assistant interface

Code

KrishTrue / dev-challange-1

hacktoberfest

AI Disease Prediction System

Advanced AI-powered disease prediction based on symptoms

An intelligent medical assistance system that uses machine learning to predict diseases based on user-selected symptoms, powered by Random Forest classification and Google Gemini AI for detailed disease descriptions.

Quick Start

For Users

Visit the web application and start predicting diseases based on your symptoms:

Pick your symptoms from 130+ options in a chat-style assistant
Get an AI-powered disease prediction
Read a structured overview: description, symptoms, causes, precautions and medication

For Developers

Prerequisites: Python 3 with uv, Node.js 20+ and pnpm

# Clone repository
git clone <repository-url>
cd dev-challenge-1

# Start backend (Terminal 1) - runs on http://127.0.0.1:5000
cd backend && ./run.sh dev

# Start frontend (Terminal 2) - runs on http://localhost:5173
cd client
pnpm install
pnpm dev

The client reads the API URL from client/.env:

VITE_API_BASE_URL=http://localhost:5000/api

System Overview

ML Model: Random Forest Classifier…

View on GitHub

Frontend: React 19, TypeScript, Vite, Tailwind CSS, Zustand
Backend & ML: Python, Flask, scikit-learn (Random Forest Classifier)
AI Synthesis: ollama pull qwen3:1.7b for detailed disease guides and precautions

How I Built It

ML Prediction Engine: Trained a Random Forest model capable of mapping 132 distinct symptoms to 41 common and acute medical conditions with ~95% accuracy.
AI Guidance Layer: Paired the prediction with AI-generated medical overviews, detailing symptoms, causes, lifestyle precautions, and common medication classes.
Chat-Style UI: Designed a friendly, clean React assistant that lets users search, select symptoms by body category, and get structured results instantly.

Why Does Open Innovation Matter?

Health literacy tools should be transparent, accessible, and community-driven. By building this with open-source tools and modular ML models, anyone can run the model locally, inspect the logic, or adapt it for local community health awareness without lock-ins or privacy compromises.

Team

Krish Puri (@KrishTrue)
Keshav Gilhotra (@ikeshav26)

DEV Community: Keshav Gilhotra