Skip to content

README.md

Latest commit

 

History

History
129 lines (82 loc) · 3.42 KB

README.md

File metadata and controls

129 lines (82 loc) · 3.42 KB

Web-Based-Leukemia-Subtype-Prediction-Using-Machine-Learning

A web-based leukemia subtype prediction tool using ML and DAA. Features include gene selection using mRMR and classification via KNN, SVM, and Random Forest. Built with Vite, React, TypeScript, Tailwind CSS (UI) and Flask (backend). Includes Knapsack for optimal feature or model selection

bio-daa.1.mp4

🧬 LeucoPredic: Web-Based Leukemia Subtype Prediction Using Machine Learning

LeucoPredic is an AI-powered web platform that predicts subtypes of Acute Lymphoblastic Leukemia (ALL) based on gene expression input. Built with a modern tech stack, it integrates machine learning algorithms and discrete algorithmic approaches (DAA) like Knapsack and mRMR for optimal feature selection and classification accuracy.

🔍 Features

  • 🧪 Predict ALL subtypes using gene expression data
  • 🧠 Machine Learning models: KNN, SVM, Random Forest, Logistic Regression
  • 🎯 Feature selection using mRMR (Minimum Redundancy Maximum Relevance)
  • 📦 Discrete Algorithmic Approaches: Knapsack for optimal model/feature optimization
  • 🌐 Modern, responsive UI built with Vite + React + TypeScript + Tailwind CSS + shadcn/ui
  • 🔗 Flask backend for model integration and inference
  • 📊 Displays prediction results with confidence scores and treatment guidance

🏥 Why This Matters

Accurate subtyping of leukemia is critical for determining the right treatment plan. This tool aids medical researchers and physicians by automating subtype prediction and reducing diagnostic errors through high-accuracy machine learning.

⚙️ Tech Stack

🧩 Frontend

  • Vite
  • React
  • TypeScript
  • Tailwind CSS
  • shadcn/ui

🔧 Backend

  • Python (Flask)
  • Scikit-learn (ML models)
  • Pandas, NumPy
  • Feature selection using skfeature or custom mRMR

🚀 Getting Started

  1. Clone the repository
git clone https://github.com/your-username/leucopredic.git
cd leucopredic
  1. Start the frontend
cd frontend
npm install
npm run dev
  1. Run the backend
cd backend
pip install -r requirements.txt
python app.py
  1. Navigate to localhost:5173 to use the app.

🧠 ML Models Used

Model Purpose
K-Nearest Neighbors (KNN) Fast, intuitive classification
Support Vector Machine (SVM) Robust decision boundary
Random Forest High-accuracy ensemble learning
Logistic Regression Baseline interpretability model

📚 How it Works

  1. User uploads gene expression data
  2. Feature selection via mRMR or Knapsack algorithm
  3. Prediction using trained ML models
  4. Result visualization with subtype classification & confidence

📂 Data Sources

  • Public leukemia datasets from NCBI GEO
  • Gene expression profiles from microarray/RNA-seq studies

✨ Sample Output

  • Subtype: B-ALL with ETV6-RUNX1
  • Confidence: 92.7%
  • Suggested Therapy: Standard chemo with excellent prognosis

👨‍💻 Author

Harish S.S. Machine Learning & Bioinformatics Enthusiast Email: harishdeepikassdeepikass@gmail.com

📄 License

This project is licensed under the MIT License. See LICENSE file for details.