Advancing Tabular Stroke Modelling Through a Novel Hybrid Architecture and Feature-Selection Synergy

2025-05-18

AI Summary

Existing stroke prediction models typically achieve accuracy below 95%, limiting their clinical utility. To address this, we propose an interpretable end-to-end framework trained on a publicly available cohort of 4,981 patients, utilizing only ten routinely collected clinical variables. Our method introduces a novel hybrid feature selection strategy combining point-biserial correlation with Random Forest Gini importance. We further design a heterogeneous ensemble architecture integrating Random Forest, XGBoost, LightGBM, and SVM, followed by logistic regression as a meta-learner—balancing interpretability and generalization. Evaluated via stratified five-fold cross-validation, the model achieves 97.2% accuracy and 97.15% F1-score, significantly outperforming the best single model (LightGBM: 91.4%). This represents near-clinical-grade performance, offering both high predictive fidelity and clinical deployability.

Technology Category

• Machine Learning: Ensemble Methods
• Natural Language Processing: Interpretability, Analysis, and Evaluation of NLP Models
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Brain stroke remains one of the principal causes of death and disability worldwide, yet most tabular-data prediction models still hover below the 95% accuracy threshold, limiting real-world utility. Addressing this gap, the present work develops and validates a completely data-driven and interpretable machine-learning framework designed to predict strokes using ten routinely gathered demographic, lifestyle, and clinical variables sourced from a public cohort of 4,981 records. We employ a detailed exploratory data analysis (EDA) to understand the dataset's structure and distribution, followed by rigorous data preprocessing, including handling missing values, outlier removal, and class imbalance correction using Synthetic Minority Over-sampling Technique (SMOTE). To streamline feature selection, point-biserial correlation and random-forest Gini importance were utilized, and ten varied algorithms-encompassing tree ensembles, boosting, kernel methods, and a multilayer neural network-were optimized using stratified five-fold cross-validation. Their predictions based on probabilities helped us build the proposed model, which included Random Forest, XGBoost, LightGBM, and a support-vector classifier, with logistic regression acting as a meta-learner. The proposed model achieved an accuracy rate of 97.2% and an F1-score of 97.15%, indicating a significant enhancement compared to the leading individual model, LightGBM, which had an accuracy of 91.4%. Our study's findings indicate that rigorous preprocessing, coupled with a diverse hybrid model, can convert low-cost tabular data into a nearly clinical-grade stroke-risk assessment tool.

Problem

• Improving stroke prediction accuracy below 95% using tabular data
• Developing hybrid model with feature selection for clinical-grade assessment
• Addressing data challenges like imbalance and missing values in stroke modeling

Innovation

• Hybrid architecture combining multiple ML algorithms
• Feature selection using correlation and Gini importance
• Data preprocessing with SMOTE for class imbalance