The rapid evolution of healthcare technology has been highly influenced by advancing machine learning approaches like federated and split learning. Federated and split learning have combined as a high standard identifying significant problems and challenges in healthcare data security and privacy. This study explores the roles of federated learning and split learning in healthcare technology. Federated learning permits models to be trained among many healthcare centres while preserving important patient data and confirming privacy. On the other hand, split learning decreases data exposure by splitting the training process and securing decentralized model training. This study reviews many challenges, problems, and applications of federated learning and split learning with machine learning and artificial intelligence in healthcare. It also summarizes the recent advances of federated and split learning in smart healthcare and describes datasets and research gaps along with comparisons of Federated Learning and Split Learning. As a result, the federated and split learning approaches have a high ability to secure patients’ important data without sharing original files, and it has a high opportunity for further research to find new techniques and technologies to preserve data in smart healthcare applications.