ML-RASPF: A Machine Learning-Based Rate-Adaptive Framework for Dynamic Resource Allocation in Smart Healthcare IoT

The growing adoption of the Internet of Things (IoT) in healthcare has led to an extensive growth in real-time data from wearable devices, medical sensors, and patient monitoring systems. This data and latency-sensitive environment poses a significant challenge to conventional cloud-centric infrastructures, which struggle with unpredictable service demands, link congestion, and end-to-end delay across distributed environments. Specifically, traditional cloud infrastructures struggle to deliver consistent service quality in smart healthcare, where both low end-to-end latency and adaptive service delivery rates are critical for life-saving applications. We propose ML-RASPF, a machine learning-based service delivery framework for efficient and scalable IoT service delivery in smart healthcare systems to address these issues. ML-RASPF formulates the provisioning task as a joint optimization problem that aims to minimize service latency and maximize delivery rate stability. The framework comprises three key components: (i) a network parameter initialization module, (ii) a supervised learning model for traffic demand prediction, and (iii) a reinforcement learning-based service adaptation engine for real-time resource allocation. These modules intelligently distribute workloads across a hybrid cloud environment. We evaluate ML-RASPF using a realistic smart hospital scenario involving IoT-enabled kiosks and wearable devices delivering both latency-sensitive and latency-tolerant services. Experimental results demonstrate that ML-RASPF significantly outperforms state-of-the-art edge–cloud systems in terms of latency reduction, energy efficiency, bandwidth utilization, and service delivery rate.