A Fog-Health Architecture for Early Alarming System of Heart Attack: A Deep Neural Network-Based Approach

In today's emerging world, technology is evolving day by day at its best. The healthcare sector is highly influenced by technological innovations. Fog computing is one of the modern technological innovations, which addresses various limitations associated with the cloud platform. This study aims to propose a fog-based healthcare architecture, as well as modules that show how Fog nodes work in this architecture. Related to this architecture, machine learning is used for real-time data analysis and future forecasts in the healthcare sector to ensure early diagnosis of heart patients. To that extent, the current study takes into account the deep neural network for classifying high-risk and low-risk patients. Thus, this research intends to implement the proposed model on the fog layer to ensure scalability and real-time data analytics. A heart disease dataset acquired from the UCI Machine Learning Datasets Repository (one of most important websites for collection of databases, domain theories, and data generators for machine learning community) is utilized for experimental purposes in this study. The results of the study show that the proposed DNN model has 82% accuracy in the classification of low-risk and high-risk patients. Moreover, it proves that the proposed model outperforms other state-of-the-art approaches, such as KNN, Decision Tree, and Random Forest by providing higher precision, recall, and F1-scores.