A Web-Based Interface That Leverages Machine Learning to Assess an Individual’s Vulnerability to Brain Stroke

Aim:

         To develop an optimized machine-learning model using Random Forest to accurately classify brain stroke risk using clinical, demographic, and physiological data.

Abstract:

        Brain stroke is a leading cause of global mortality and long-term disability, making early detection essential. Traditional assessment methods rely on manual evaluation and simplified scoring, which cannot analyze complex clinical interactions. This project introduces a data-centric machine learning framework using Random Forest, supported by structured preprocessing, label encoding, and exploratory data analysis. The system effectively models nonlinear risk patterns and provides interpretable predictions. Results demonstrate strong accuracy and reliability, highlighting its value for clinical decision support.

Proposed System:

      The proposed system employs a Random Forest classifier combined with advanced preprocessing, label encoding, and exploratory data analysis. It identifies significant predictors such as age, hypertension, glucose levels, BMI, and smoking status. Hyperparameter tuning enhances model accuracy, while feature-importance insights improve transparency. Integrated into clinical workflows, the system provides fast and reliable stroke-risk predictions.

Advantages:

• High accuracy and robustness.
  • Handles imbalanced and noisy clinical data.
  • Provides interpretable feature-importance insights.
  • Scalable for real clinical environments.
  • Supports early diagnosis and intervention.
  • Resistant to overfitting due to ensemble structure.