Balancing and metaheuristic techniques for improving machine learning models in brain stroke prediction - Research summary

Aug 4, 2026

Upload by Eden

Stroke is a time-critical brain condition caused by disrupted blood flow, making early detection essential to reduce fatal complications and long-term disability. The study addresses an imbalanced stroke dataset of 4,981 records by applying KMeansSMOTE for class balancing. Five learning models—decision tree, random forest, SVM, K-nearest neighbors, and gradient boosting—are further improved through hyperparameter optimization using four metaheuristics: GWO, PSO, GA, and ABC. Evaluation uses accuracy, recall, precision, F1-score, and AUC. Random forest tuned with genetic algorithm achieves the best results, highlighting the impact of balancing plus metaheuristic optimization on stroke prediction.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages9

Reading Time23 min

AccessFree to download

Article Info ABSTRACT

Introduction

Related Work and Motivation

Balancing and metaheuristic techniques for improving machine learning models in brain stroke prediction Abd Allah Aouragh1, Mohamed Bahaj1, Fouad Toufik2 1MIET Laboratory, Faculty of Sciences and Techniques, Hassan 1st University, Settat, Morocco 2Computer Sciences Laboratory, Higher School of Technology, Mohammed V University, Sale, Morocco | Article history:
Received Mar 19, 2024 Revised Oct 19, 2024 Accepted Oct 23, 2024 | A brain stroke, medically referred to as a stroke, represents a critical condition triggered by the disruption of blood flow to a region of the brain. Early detection of stroke is crucial to prevent fatal complications. In this study, we worked with an unbalanced dataset of 4981 entries on stroke, which we balanced using the K-means synthetic minority over-sampling technique (KMeansSMOTE) algorithm. We then employed five machine learning algorithms: decision tree, random forest, support vector machine, K-nearest neighbors, and gradient boosting. We compared the hyperparameter optimization of these algorithms using four metaheuristic techniques: gray wolf optimization, particle swarm optimization, genetic algorithm, and artificial bee colony. The models' effectiveness was evaluated using multiple metrics, such as accuracy, recall, precision, F1-score, and area under the receiver operating characteristic curve. Our findings indicate that the random forest optimized by the genetic algorithm achieved the best performance, with an accuracy of 97.39% and an F1-score of 97.35% . This study highlights the effectiveness of balancing and metaheuristics techniques in optimizing machine learning models for stroke forecasting.
This is an open access article under the CC BY-SA license.
| | --- | --- | | Keywords:
Brain stroke
Genetic algorithm Hyperparameter optimization KMeansSMOTE
Machine learning Oversampling
Random forest | | Corresponding Author: Abd Allah Aouragh MIET Laboratory, Faculty of Sciences and Techniques, Hassan 1st University Settat, Morocco Email: [abdallahaouragh@gmail.com](abdallahaouragh@gmail.com) Article Info ABSTRACT 1. INTRODUCTION Brain stroke, also known as stroke, is a critical health problem worldwide, constituting one of the foremost causes of mortality and long-term handicap [1] . When a stroke occurs, the risk of death is high. If not fatal, a stroke can cause vision or speech impairment, paralysis, and confusion. Every year, 15 million people are reported to have a stroke: 5 million die, and 5 million are left permanently disabled, burdening families and societies [1], [2] . A stroke happens when blood flow to a section of the brain is disrupted, leading to oxygen deprivation, and subsequently, tissue damage. Prompt recognition and early intervention are paramount in mitigating the devastating consequences of this condition [3] . While there are many treatments available for stroke, including surgery, radiation therapy, chemotherapy, and targeted therapeutic approaches [4], these interventions can be costly, and their effectiveness often depends on how quickly the stroke is diagnosed and treated. Therefore, early diagnosis of stroke is of utmost importance, as it not only improves the health situation of patients but also reduces the costs associated with their rehabilitation [5] . Although potentially serious, stroke remains a challenging condition to predict and manage effectively [3], [4] . Machine learning techniques have become promising instruments in the healthcare field, facilitating the analysis of complex medical data and aiding clinical decision-making [6] . These advancements have also improved the prediction of various diseases, including brain tumors [7], liver disease [8], and others [9], notably through the utilization of dataset balancing and hyperparameter optimization techniques [10], [11] . In the context of stroke, the utilization of machine learning algorithms and diverse optimization techniques for early prediction ho

FAQs

How does the study handle class imbalance in the stroke dataset?

It balances an imbalanced dataset of 4,981 entries using the KMeansSMOTE synthetic minority over-sampling technique before training models.

Which machine learning algorithms are compared in the experiments?

The experiments compare decision tree, random forest, support vector machine, K-nearest neighbors, and gradient boosting.

What metaheuristic methods are used for hyperparameter optimization, and what is the best result?

Four techniques are used: gray wolf optimization, particle swarm optimization, genetic algorithm, and artificial bee colony. The best performance is achieved by a random forest optimized with the genetic algorithm, reaching about 97.39% accuracy and 97.35% F1-score.

Aug 4, 2026

Upload by Eden