The paper introduces the FedHealthFog framework, a federated learning (FL) approach designed for healthcare analytics over a fog computing platform. The framework aims to address the challenges of distributed learning in resource-constrained IoT-enabled healthcare systems, particularly those sensitive to delays and energy efficiency. Traditional FL approaches often suffer from high computational requirements and communication costs due to their reliance on a central server for data aggregation. FedHealthFog overcomes these issues by elevating fog nodes to the role of local aggregators within the FL architecture. A greedy heuristic technique is used to optimize the selection of fog nodes as global aggregators, reducing communication latency and energy consumption. The system demonstrates significant improvements in latency reduction (up to 87.01% for the test set) and energy efficiency (up to 57.98% for the training set) compared to benchmark algorithms. The effectiveness of FedHealthFog is validated through experiments, showing its potential to transform FL in resource-constrained IoT environments for delay-sensitive applications.The paper introduces the FedHealthFog framework, a federated learning (FL) approach designed for healthcare analytics over a fog computing platform. The framework aims to address the challenges of distributed learning in resource-constrained IoT-enabled healthcare systems, particularly those sensitive to delays and energy efficiency. Traditional FL approaches often suffer from high computational requirements and communication costs due to their reliance on a central server for data aggregation. FedHealthFog overcomes these issues by elevating fog nodes to the role of local aggregators within the FL architecture. A greedy heuristic technique is used to optimize the selection of fog nodes as global aggregators, reducing communication latency and energy consumption. The system demonstrates significant improvements in latency reduction (up to 87.01% for the test set) and energy efficiency (up to 57.98% for the training set) compared to benchmark algorithms. The effectiveness of FedHealthFog is validated through experiments, showing its potential to transform FL in resource-constrained IoT environments for delay-sensitive applications.