Diagnosing Depression with Artificial Intelligence: Systematic Literature Review

Background: Artificial intelligence is a new wonder to predict and assess the severity of mental health disorders. Undoubtedly, the Audio/Visual Emotion Challenge (AVEC 2016) has boosted the research in the related subject but there is still a dearth of reviews related to artificial intelligence and depression. Aim: To extrapolate the handouts about the quality assessment of included studies of this Systematic Literature Review (SLR). To find out the most widely adopted AI approach to predict depression. To find out the most widely used biomarker for predicting depression. To identify the Model Accuracy of the included studies. Methods: The articles related to the applications of artificial intelligence predicting depression were searched by inputting the keywords (“Depression”) AND (“Artificial Intelligence” OR “Machine Learning” OR “ML”) AND (“Prediction” OR “Diagnosis”) in the search engine of websites like Science Direct, Web of Science, PubMed and Springer link. The outputs were filtered with the inclusion and exclusion criteria. The quality assessment was done based on the Quantitative Assessment Tool for Studies with Diverse Designs (QATSDD). Results: A narrative approach was adopted to encapsulate the Quality assessment of included studies to predict depression. The findings of this study concluded that the study carried out by Marques, et al., (2020) came with the highest quality score of 36 on rating with the QATSDD among the included studies of this Systematic Literature Review (SLR). This study also found that the top accuracy of the developed model was detected from the model developed by Sharma, et al., (2018).