Internet of Medical Things (IoMT) are networks which are targeted towards design of healthcare communication interfaces with low latency and high security. In order to design such interfaces, efficient models for data encryption, hashing, privacy, and quality of service (QoS) awareness are needed. A wide variety of standard medical interfaces are proposed by researchers, which assist in reducing network redundancies for high-throughput and low latency communications. These interfaces also implement security models that ensure data encryption & privacy. But due to incorporation of encryption methods, QoS performance of the IoMT devices reduces, which limits their real-time usability for in-patient monitoring & treatment. In order to improve IoMT QoS while maintaining high security, this text proposes design of QSIH, which is a QoS-aware sidechain model that can be used for securing IoMT networks. The proposed model describes design of a blockchain-based data storage & communication interface, which is capable of removing a wide variety of network attacks. The delay needed for communication in any blockchain-based interface increases exponentially w.r.t. number of blocks added to the system. In order to reduce this delay, a novel machine learning model based on Genetic Algorithm optimization is proposed. The proposed model splits the main blockchain into multiple shards in a QoS-aware manner, thereby ensuring low delay, and high communication throughput. The shards (or sidechains) are managed using an interactive Q-Learning (IQL), which is able to expand or contract these chains depending upon network’s QoS performance. Sidechains which are unused for large periods of time are combined together, and archived for future reference. The archived sidechains are formed from main blockchain, and are merged with other sidechains depending upon archival requirements of the network. Due to such a dynamic side chaining model, the proposed QSIH model is capable of reducing network communication delay by 18%, increase throughput by 14%, reduce storage cost by 5%, while maintaining high level of security & privacy in the network. The model was tested under different IoMT scenarios, and it was observed that it showcased consistent performance across different network emulations.