Empowering heart attack treatment for women through machine learning and optimization techniques

Area of Science:

• Cardiovascular Medicine
• Artificial Intelligence
• Machine Learning

Background:

• Heart attack diagnosis in women is often delayed or incorrect due to unique symptom presentation and physiology.
• Existing diagnostic models require optimization to address sex-specific differences in cardiovascular disease.

Purpose of the Study:

• To develop and validate an optimized ensemble learning model for enhanced heart attack classification accuracy in women.
• To integrate novel optimization algorithms for improved diagnostic performance in cardiovascular care.

Main Methods:

• An ensemble learning approach using a voting classifier combining the Waterwheel Plant Algorithm (WWPA) and Stochastic Fractal Search (SFS).
• Integration of multiple machine learning classifiers (Gaussian Naive Bayes, Random Forest, Logistic Regression, SGD, SVC, Decision Tree, k-NN).
• Evaluation on a clinical dataset of 303 patients with 14 features, using 10-fold cross-validation, ANOVA, and Wilcoxon Signed Rank Test.

Main Results:

• The proposed WWPA+SFS model achieved a highest classification accuracy of 97.01%.
• The model demonstrated robust performance and low variance across multiple trials.
• Outperformed other optimization algorithms including GWO, WOA, PSO, and GA.

Conclusions:

• The WWPA+SFS ensemble learning model significantly optimizes heart attack detection accuracy in women.
• This approach holds potential for reducing misdiagnosis rates, lowering healthcare costs, and advancing personalized cardiovascular treatment.