Intelligent Methods for Early Prediction of Heart Disease

• Published in: International Conference on Control, Decision and Information Technologies (Online) pp. 2574 - 2578
• Main Author: Al-Jamimi, Hamdi A.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 03.07.2023
• Subjects: Heart; Machine learning algorithms; Medical diagnosis; Medical diagnostic imaging; Minerals; Prediction algorithms; Training
• ISSN: 2576-3555
• DOI: 10.1109/CoDIT58514.2023.10284226

The early diagnosis of heart disease relies on accurately assessing symptoms that manifest at different stages of the disease. The health care sector holds extensive patient records containing invaluable data that necessitate in-depth investigation for uncovering hidden patterns crucial to early disease diagnosis. Timely detection of cardiovascular disease has the potential to save lives. Consequently, the application of intelligent methods in biomedical research has garnered significant interest, leading to the exploration of various artificial intelligence techniques at different levels within the medical domain. These techniques aim to utilize the existing data to improve early detection and mitigate the risks of numerous diseases, including heart disease. This study highlights the importance of leveraging machine learning (ML) methods for accurate heart disease prediction. By utilizing a diverse range of ML algorithms and a carefully collected dataset, we have successfully developed and evaluated multiple classifiers for heart disease diagnosis. The performance of these classifiers is evaluated using a newly published dataset that was collected from the Medical Help Center, a private hospital and heart center in Erbil, Iraq. This dataset comprises 333 cases and 21 attributes, which are utilized for training and validating the developed ML models. Experimental results demonstrate that the employed techniques achieve an accuracy level of over 92%. Notably, the gradient-boosting classifier exhibits the highest performance, achieving an accuracy level of 98.51 % during the testing phase. The promising results obtained demonstrated the potential of ML techniques in enhancing early detection and improving patient outcomes.