D-Khalid/GeneScout_Predtive_AI_Pathologist

🧬 GeneScout: AI-Powered Genetic Disease Pathologist

🚀 Project Overview

GeneScout is an interpretable Machine Learning diagnostic tool designed to predict 5 genetic diseases (Cystic Fibrosis, Sickle Cell, etc.) based on patient biomarkers. Unlike "Black Box" models, GeneScout prioritizes clinical explainability using SHAP values.

🛠️ Tech Stack

• Model: Voting Ensemble (Random Forest + SVM + Logistic Regression)
• Explainability: SHAP (Shapley Additive exPlanations) for global and local feature importance.
• Deployment: Streamlit Web App for real-time inference.
• Data Analysis: Pandas, Seaborn, Matplotlib.

📊 Key Findings

1. Sweat Chloride was identified as the primary biomarker for Cystic Fibrosis (SHAP value > 0.8).
2. Hemoglobin & Fetal Hemoglobin levels successfully differentiated Thalassemia from anemia.
3. Accuracy: The ensemble model achieved 93.5% Accuracy on the test set.

📂 Try out the Web-app

streamlit

(https://genescout-ai-genetic-disease-pathologist-sbqf5qzusuptbdwrslbmn.streamlit.app/)

HuggingFace Space

(https://huggingface.co/spaces/D-Khalid/GeneScout_AI_Predictive_Pathologist)

📂 Project Structure

• app.py: The Streamlit dashboard.
• train_model.py: Training script for the Voting Classifier.
• explain_model.py: SHAP analysis and plot generation.
• eda_analysis.py: Initial data exploration.