Detection of Depression Using Data Collected from Social Media

In recent times, detecting depression using data collected from social network, like audio, video, image or text data, is a growing research field in emotion processing. Data from social media, which includes text data and gifs, emojis, etc., would be used for the analysis and classification processes. An autonomous depression detection system is required due to the complexity of conventional methods based on clinical diagnosis. Using data collected from social media, sentiment levels can be identified and monitored. In this paper, we aim to research on using machine learning/deep learning techniques combined to make a hybrid model. CNN is a technique proven to be efficient in feature extraction combined with Bidirectional LSTM, for contextual learning, both used to apply on the data collected from social media, train the data and obtain efficient results on the level of depressional traits in an individual. It has also been possible to get a reasonably high value of accuracy of 93% upon evaluation. The use of AI and machine learning in such systems has greatly assisted in many fields of medical science including psychology analysis.