Depression Identification Through Tweet Clusters

Over the past few years, the awareness and popularity of Mental health have been on a rapid rise and people are becoming more aware of the surrounding problems. This has helped for mental illnesses like Depression to become recognized and be treated appropriately. Social media has played an integral part in this uproar due to its increased popularity and ease of use. This has allowed people to spread awareness, seek help and vent out their emotions. Our paper is a comparative study of different models for detecting depression with real-time Twitter data and proposing the best performing model. For depression detection, a collection of tweets per user spread over time was used. The data was augmented and then passed through the deep learning model to identify depression in Twitter users based on their Time-Distributed tweets. The proposed model achieved an accuracy of over 90%.