Abstract:
Cardiovascular diseases are one of the leading causes of death worldwide and anemia is a common
condition marked by low hemoglobin levels. Despite its prevalence anemia is often overlooked even though it
plays a significant role in increasing the risk of heart disease. This study aims to develop a predictive model to
assess the heart disease risk in patients with anemia. For this research, hematological and cardiovascular data
were collected from 800 Bed Head Tickets at Jaffna Teaching Hospital, Sri Lanka. The data included patients
with heart failure, myocardial infarction and those without heart disease. Manual data collection is difficult
and takes time so MIMIC database was chosen as a backup. It is a publicly available dataset containing
deidentified patient details and supported the initial stages of model implementation. After completing the
manual data collection we applied multiple machine learning algorithms along with two ensemble approaches
to train the data. The ensemble combining Logistic Regression and Support Vector Machine (SVM) achieved
the highest accuracy of 79.38% and ROC-AUC of 82.02% with strong performance in other metrics. The
proposed model demonstrates strong potential for clinical application by enabling early detection of high risk
anemia patients. This will help facilitate timely medical check ups and reduce the risk of heart complications.
