Heart Failure Prediction Models using Big Data Techniques

Big Data technologies have a great potential in transforming healthcare, as they have revolutionized other industries. In addition to reducing the cost, they could save millions of lives and improve patient outcomes. Heart Failure (HF) is the leading death cause disease, both nationally and internally. The Social and individual burden of this disease can be reduced by its early detection. However, the signs and symptoms of HF in the early stages are not clear, so it is relatively difficult to prevent or predict it. The main objective of this research is to propose a model to predict patients with HF using a multi-structure dataset integrated from various resources. The underpinning of our proposed model relies on studying the current analytical techniques that support heart failure prediction, and then build an integrated model based on Big Data technologies using WEKA analytics tool. To achieve this, we extracted different important factors of heart failure from King Saud Medical City (KSUMC) system, Saudi Arabia, which are available in structured, semi-structured and unstructured format. Unfortunately, a lot of information is buried in unstructured data format. We applied some pre-processing techniques to enhance the parameters and integrate different data sources in Hadoop Distributed File System (HDFS) using distributed-WEKA-spark package. Then, we applied data-mining algorithms to discover patterns in the dataset to predict heart risks and causes. Finally, the analyzed report is stored and distributed to get the insight needed from the prediction. Our proposed model achieved an accuracy and Area under the Curve (AUC) of 93.75% and 94.3%, respectively.