Disease diagnosis using machine learning and deep learning models has become a hot research topic in the medical era. Detection and diagnosis of diseases using deep learning models enhance quick decision-making and reduce the intervention in its severity. Several machine learning and deep learning models greatly impact the development of the detection and diagnosis of several diseases. The adopted supervised deep learning model for disease diagnosis is ineffective due to the unavailability of a large number of labeled samples for training and classification. Existing deep learning models have resolved the supervised issue, but the potential measure of unsupervised learning is quite crucial. Hence, to design unsupervised deep learning methods in existing deep learning models for disease diagnosis is essential to define the appropriate class labels of the input samples and for proficient disease prediction.