Investigation of the effect of COVID-19 on heart attack risk using machine learning techniques

The size of data is rapidly increasing day by day. Uncovering and leveraging valuable information embedded within large volumes of data is becoming increasingly crucial in today's competitive world. When dealing with such large volumes of data, especially the challenge of how to reliably analyze these data has also emerged. At this point, data mining has facilitated researchers in the process of analyzing large datasets. Healthcare services detailed processes for the diagnosis, treatment, and prevention of diseases, injuries, and other physical and mental disorders in individuals. The healthcare sector can be considered a rich source of data due to the vast amount of information it generates, including electronic medical records, administrative reports, and other comparative findings. In healthcare service, data mining is primarily used to aid in predicting various diseases and assisting doctors in making clinical decisions, as well as in diagnosis and prognosis. In this thesis study, it is aimed to diagnose heart attack by using a data set enriched with attributes of COVID-19. The diagnosis of heart attacks was made using ensemble learning algorithms including Random Forest, XGBoost, LightGBM, and CatBoost, with a dataset comprising 205 patients, including qualitative fields related to COVID-19, collected in collaboration with the Health Sciences University Dr. Siyami Ersek Thoracic Cardiovascular Surgery Training and Research Hospital. According to the recall measure, the algorithm with the best performance is CatBoost, followed by the Random Forest algorithm. However, according to the F1-measure, the Random Forest algorithm exhibits the best performance.