In the dynamic environment of health and information and communication technology (ICT), it has served as a guide and has significantly affected all levels of the healthcare model. A combination of health information technology and information technology medicines uses computer hardware and software to collect, process, store, distribute, exchange information and make decisions in electronic form of audio files, images, Responsible for sharing texts and comprehensive digital information. The rapid development of technology and health information encourages various institutions to provide medical services to obtain the skills needed to provide highly qualified medical services. Electronic health records (EHR) and electronic medical records (EMR) are one of the most widely used technologies in health systems, and in addition, much research has been done on this topic in recent years. Electronic medical records and electronic medical records contain all information related to the health of citizens even before they were born after their death. It is continuously collected over time and stored electronically. All or part of the record can be granted by the authority holder, regardless of location or time. The electronic medical record EMR represents one of the most ideal medical information systems (MIS), which must be carefully designed and managed to meet the needs of society. The introduction and deployment of electronic medical records and electronic medical records is the ultimate goal of establishing health system information technology, but this always involves many obstacles and challenges. Using different perspectives, researchers have considered obstacles, acceptance, use, continued use, and implementation, operation, and implementation of electronic medical record systems, and have achieved a wealth of results. Machine learning methods have been adopted for diagnosis and disease prediction, but some challenges remain. Dataset quality is one of the key factors for supervised learning and unsupervised learning models. Models trained on incorrect or biased data can lead to poor quality predictions. Machine learning models require sufficient training data to have good classification and prediction results. However, medical data extracted from electronic medical records and electronic medical records often suffers from data loss issues for a variety of reasons, including testing equipment, disease progression, and other availability.