Artificial Intelligence (AI) and the Internet of Things (IoT) together have the potential to completely transform healthcare in Society 5.0, especially with Federated Learning (FL). The Internet of Medical Things (IoMT) will heavily rely on FL, a subset of AI, to handle massive volumes of data securely and protect patient privacy. FL functions in a decentralized manner as opposed to centralized methods, protecting private information from being shared with a central system by training local models on individual nodes. Establishing confidence among participants is crucial in this collaborative approach to safeguard the integrity of the FL global model. The evolution of the centralized model could be jeopardized by any malevolent node action, underscoring the importance of cooperation and mutual confidence. FL not only makes medical innovation and image analysis more efficient but it also optimizes time and money spent on healthcare procedures. As of right now, FL comes in three flavors: Federated Transfer Learning, Vertical FL, and Horizontal FL. Each has a distinct function in the IoMT environment. To create effective gradient synchronization protocols in IoMT contexts, however, focus needs to be placed on improving the security and privacy of FL models, especially with relation to network communication. In the long run, integrating FL models with e-Healthcare systems is a big step toward utilizing AI-powered IoT technologies to improve healthcare. In the future, according to Society 5.0, the seamless integration of AI and IoT in healthcare would prioritize data security and privacy while also improving patient outcomes in the digital age.