Wearable health devices have transitioned to real-time monitoring and prediction of cardiovascular health, enabling early detection and prevention of such diseases. This research proposes an energy-efficient management framework for real time cardiovascular disease prediction through an IoT-Fog-Cloud architecture by moving all data processes and computations to the Fog node. The objective is to develop a health monitoring model using a real-life dataset from Federal Teaching Hospital Gombe State, Nigeria. The model employs machine learning models and deep learning for prediction and feature selection. The hybrid strength of Naive Bayes (NB) and Random Forest (RF) was used for prediction and optimised using the Particle Swarm Optimisation Technique(PSO) to achieve computational accuracy and energy efficiency. The ONBRF was evaluated using accuracy, latency and energy consumption to achieve (92.3%) accuracy, 90.4Wms of consumed energy and 301ms of latency. The experiment used R programming language and deployed in real time on a web R shiny cloud environment. This study was specifically designed for cardiovascular disease predictions. An elaborated methodology can be implemented in future studies for other disease predictions.