Anxiety, depression, and stress are all psychiatric disorders that coexist and have an impact on the quality of life of people across the globe. They are influenced by environmental, psychological, and biological factors. Prognosis is important for early treatment and to minimize the effects of these diseases on the individual as well as on society. According to the WHO estimate, 1 out of every 8 people in the world have a mental illness. Severe impairment in the thought, emotional, or behavioural processes is a characteristic of mental diseases. Usually, wearable technology, social media activity, or self-reported questionnaires are used to gather data for stress, anxiety, and depression prediction. In order to identify patterns and risk factors for accurate prediction we generally use machine learning models and statistical techniques. The outcomes portray moderate to high effectiveness in depression, anxiety, and stress prediction, varying with the methodology and data quality. We have reviewed 13 different models for the prediction of stress, anxiety and depression. By comprehensive study we find out that neural network performed the best with the highest accuracy in terms of Accuracy, Error rate, Precision, Recall, F-measurer area.