Using Ensemble Deep Learning and Feature Engineering Approaches to Classify Hospital Readmitted Heart Disease Patients

Heart disease is one of the leading causes of death worldwide and the costs associated with treating heart disease patients in the US is estimated to surpass $1 trillion by 2035. This is mainly due to the high rate of hospitalization of heart disease patients, as well as their high rate of readmission after being discharged from the hospital. In addition, currently the diagnosis of heart disease is primarily done by medical experts based on their experience and understanding of the disease and after reviewing patients' health history and clinical data, which increases the risk of errors, treatment time, and eventually the total cost of treating heart disease patients. To increase the efficiency of treatment offered by the healthcare professional and reduce the associated costs, it is crucial for an automated prediction model to be developed that can help reduce readmission rates by identifying high-risk patients before they are discharged, and accurately identify at-risk patients who have not yet been diagnosed with heart disease. This study combines deep learning and feature engineering approaches, and proposes a novel stacking ensemble deep learning model to classify hospital readmitted heart disease patients. The proposed model is evaluated by comparing its performance with various general and deep learning models using multiple performance evaluation metrics. The results demonstrate that the proposed model achieves the highest prediction performance among all metrics. Additionally, the study proposes how this model can be realistically deployed in real-world to augment critical healthcare professionals.