A Machine Learning Framework For Early Stroke Detection Using Genetic Algorithms And Bilstm Networks

Abstract

Cerebrovascular diseases, particularly stroke, are among the leading causes of death and long-term disability
worldwide; however, early diagnosis and timely intervention can significantly reduce their impact and
improve clinical outcomes. In recent years, machine learning techniques have gained considerable attention
for their potential to support early stroke detection. This study aims to identify reliable algorithms, features,
and methods that can assist medical professionals in making informed decisions for stroke diagnosis and
prevention. To achieve this objective, an early stroke detection system is proposed using brain CT images
combined with a genetic algorithm and a bidirectional long short-term memory (BiLSTM) network. A neuralnetwork-
based genetic approach is employed to select the most relevant features for accurate classification,
which are then provided as input to the BiLSTM model for prediction. The system performance was evaluated
using cross-validation and multiple metrics, including accuracy, precision, recall, F1-score, ROC curve, and
area under the curve (AUC), to ensure comprehensive assessment. Experimental results demonstrate that the
proposed diagnostic framework achieves an accuracy of 97%, outperforming conventional machine learning
models such as Logistic Regression, Decision Trees, Random Forests, Naive Bayes, and Support Vector
Machines. The developed system provides effective decision support for clinicians, enabling more reliable
and early stroke diagnosis.