Heart Failure Identification and Classification using Unstructured Dataset of Cardiac Patients

Abstract

Heart Failure (HF) has become a major health problem throughout the world. Its occurrences increase with the age of patients, and it has become the main cause of the high mortality rate in elders. A successful progression and identification of HF can be very useful to lessen the social and individual problem from this syndrome. A lot of raw clinical data are available in health care institutions in the form of patients’ medical reports, electronic test results and medication history. There is a need to convert this data set in electronic form and to get hidden information and patterns. It would help the medical practitioners to make earlier intelligent decisions about the risks of HF. Data mining techniques have great potential to extract these hidden information and patterns from such data set. This research study contains a real data set of cardiac disease patients from a renowned cardiology hospital in Pakistan to develop an HF identification and classification model using this data set. This data set has divided into three groups according to the patient’s age, namely young, adult, and old and then further classified each age group into four classes according to present physical situation of the patients, namely normal, low risk, high risk, and critical. Latest data mining algorithms have applied to each separate class of every age group to identify and classify the HF patients. The results of this study show that Decision Tree (DT) gives the highest accuracy result of 90% and outperform all other state-of-the-art algorithms. Our proposed model correctly identifies various stages of cardiac patients for each age group and it can be very beneficial for the early detection and prediction of HF risk factors. This study also provides a summary of modern strategies for treatment of HF patients for each physical class that have appeared in the past few years.