🧠 Stroke Prediction using Machine Learning

A data-driven healthcare project that predicts the likelihood of a stroke based on patient health metrics. Designed for early diagnosis support and medical decision-making.

🚀 Project Overview

This ML-based system analyzes patient records to determine stroke risk with a focus on accuracy and clinical relevance. The project applies statistical analysis, advanced preprocessing, and robust classification techniques.

📌 Features

• Cleaned and validated real-world healthcare dataset
• Balanced class distribution using resampling techniques
• Applied models: Logistic Regression, Random Forest, XGBoost
• Performance metrics: Accuracy, F1-score, ROC-AUC, Confusion Matrix
• Feature importance insights to support interpretability

🗃️ Dataset Source

Kaggle: Stroke Prediction Dataset
Fields include:

• age, hypertension, heart_disease
• avg_glucose_level, bmi
• gender, smoking_status, work_type, ever_married

✅ Results Summary

• Best model: Random Forest
• ROC-AUC: ~0.92
• Precision-Recall optimized for minimizing false negatives (critical in healthcare)

💡 Applications

• Virtual health assistants
• Risk flagging for hospitals and insurance firms
• Embedded ML in medical dashboards

---

🛑 Disclaimer: This is a research and demo project only. It is not validated for clinical use.