Machine learning (ML) based optimization algorithms have been applied in Wireless Body Area Networks (WBANs) for IoT health care to improve network performance. These algorithms can be used for various purposes, such as Channel allocation, Quality of service, Energy optimization, and Fault tolerance. Using a Q-learning algorithm in WBANs can help improve the accuracy and efficiency of IoT healthcare systems, leading to better patient outcomes. The learning rate of the Q-learning is enhanced by utilizing the Adagrad ALR optimizer. Q-learning with Adagrad ALR optimizer-based channel allocation can be used to optimize channel allocation by considering factors such as network congestion, link quality, and node power constraints by optimizing channel allocation. It will improve the performance of WBANs, leading to faster and more reliable medical data transmission. The proposed Q-learning with Adagrad ALR optimizer algorithm dynamically adjusts channel allocation in real-time based on changing network conditions, leading to more efficient use of available channels. In addition to improving network performance, ALR-based channel allocation can help extend battery life and reduce energy consumption in WBANs. By optimizing the use of available channels dynamically, ALR algorithms can help reduce the amount of energy consumed by the network, leading to longer battery life and reduced costs associated with IoT healthcare systems. To validate the performance of the proposed Q-learning with the Adagrad ALR optimizer method, the simulation results were compared with the three existing channel allocation mechanisms such as the Q-learning method, PEH quality of service, and the Clustering algorithm in terms of throughput, delay, and energy efficiency. The energy efficiency of the proposed algorithm gets enhanced by 17% when compared with the other three algorithms.