A Comparative Study on Machine Learning Techniques in Prediction of Cardiovascular Diseases

Background: The leading rank of cardiovascular diseases (CVDs) in worldwide mortality urges the critical need for predictive tools to enable prevention and early intervention. With the advent of large-scale epidemiological datasets and advancements in computational techniques, machine learning (ML) has emerged as a transformative approach for CVDs risk prediction. Materials and Methods: This study utilizes data on ۱۵۰۰۱ individuals from the Azar cohort in Iran, to evaluate the effectiveness of ML algorithms, including Support Vector Machines (SVM), Random Forests (RF), XGBoost, and Artificial Neural Networks (ANNs), in predicting CVDs. Hyperparameter tuning via grid search was conducted to optimize model performance. Results: The results demonstrate that RF and XGBoost achieved the highest accuracy (۹۴%), followed closely by ANN (۹۳%) and SVM (۹۱%). Conclusion: These findings underscore the potential of ML in enhancing CVDs risk prediction and inform future research aimed at refining predictive models through hybrid algorithmic approaches.

کلیدواژه ها:

Artificial neural network ، CVD ، random forest ، Support vector machine