A substantial amount of data related to various diseases was collected every year from different medical universities and hospitals worldwide, which was utilized to assess disease rates manually. However, it had not been adequately harnessed to establish connections between symptoms and disease risk. Machine learning (ML) and deep learning (DL) had become popular as technologies that were considered more precise and efficient in a variety of medical issues, including diagnosis, prognosis, and intervention. These were representational learning techniques that were used to nonlinearly transform the data, revealing hierarchical connections and patterns. To create effective methods for reducing the various risk factors of different diseases, it was necessary to properly understand and critically analyze the current ML and DL models. This work provided a cogent assessment of the shortcomings of the existing systems and covered the growing corpus of recent literature on ML and DL models for predicting various diseases. For an assessment of the state-of-the-art, the taxonomic structure of the available literature on predicting various diseases was examined, broken down into the techniques employed, projected outcomes, factors involved, types of datasets used, and corresponding goals.