Scientists have developed a new algorithm that can spot depression in Twitter users. The computations of the algorithm are 90 percent accurate and may lead to future early diagnosis methods. It is capable of determining Twitter users’ mental health and analyzing 38 data points from their public profile, this includes content from their posts, post timings, and other users in their social circle.
Many potential people suffering from depression across the world do not seek professional help, due to various factors. These include social stigma or themselves being unaware of their mental condition, leading to “severe delay of diagnosis and treatment”.
Abdul Sadka, director of Brunel’s Institute of Digital Futures, study co-author in a statement said,
“We tested the algorithm on two large databases and benchmarked our results against other depression detection techniques. In all cases, we’ve managed to outperform existing techniques in terms of their classification accuracy.”
Earlier research shows, that social media data can offer some valuable clues on the physical and mental health status of individuals.
According to the latest study, the algorithm is trained to use two databases, it contains Twitter history of thousands of users, alongside additional information about those users’ mental health. The researchers used 80% of the information from each database to teach the bot, and the remaining data to test its accuracy.
Abdul Sadka, further added, “In this paper, we argue that it is feasible to identify depression at an early stage by mining online social behaviors.”
The bot runs through the database and excludes users with less than five tweets and does corrections for misspellings and later considers 38 distinct factors – such as a user’s use of positive and negative words, the number of friends and followers they have, and their use of emojis – to estimate a user’s mental and emotional state. The results managed to achieve 89 percent according to the team.
The researchers were able to achieve around 71% accuracy using John Hopkins University’s CLPsych 2015 dataset.
Dr. Sadka said, “It’s not 100% accurate, but I don’t think at this level any machine learning solution can achieve 100% reliability. However, the closer you get to the 90% figure, the better.”
According to the researchers, the bot can potentially flag a user’s depression before they post something in the public domain. This may pave a way for social media platforms like Twitter and Facebook to proactively flag mental health concerns with users.
Finally, the bot can also be used for a number of applications such as sentiment analysis and criminal investigations in the future. Though the findings of the research raise questions about data privacy and require informed consent from users before their public data is used for analysis.