Through the hasty growth of data generated by intelligent IoT devices, Federated learning (FL) seems to be a promising technique that provides distributed Machine Learning (ML) amenities at the same time protecting data privacy. FL is the novel form of Artificial intelligence (AI) that builds on decentralized data setup and carries out training that brings learning to devices. It’s mostly used in instances that involve security and privacy as the main concerns and empowers implementers to build secure learning environments. The federated averaging (FedAvg) is one the most used optimization algorithms that train models with a synchronized protocol. However, the algorithm is not realistic enough and communication efficiency issues tend to arise. The amount and distribution of collected data have a different training process because of varying sample sizes of devices. This paper carries out an in-depth review of FL and its asynchronous learning previous research. Lastly, the authors propose a privacy-preserving asynchronous FL framework for distributed healthcare care data that improve the model accuracy to health information. Although the framework is still being implemented it aims at guaranteeing improved communication amongst healthcare industry participants such as hospitals, clinics, laboratories, pharmaceuticals, and many more facilities.