How Twitter can detect users with depression and anxiety
| May 17, 2019, 11:51 IST
A new study reveals how Twitter can be used to detect users with depression and anxiety. People who suffer from these mental health issues are likely to post photos on the social media site with less vivid colours and low aesthetic values, according to the researchers. Researchers hope Twitter can be used as a way to screen for these conditions.
Using algorithms, the team was also able to extract specific features like facial expressions and colours from more photos posted by more than 4,000 users.They also analysed each user's tweets and close to 887 tweeters were given a survey so researchers could score their levels of depression and anxiety. These users are also more likely to hide their negative emotions and suppress positive ones, according to the study's findings. Exhibiting a straight face in their profile pictures is one example. The results also revealed users with anxiety were more likely to have images with grayscale and low aesthetic cohesion. Meanwhile, depressed users were found to post photos only of themselves alone. These users also did not post photos that showed recreational activities or their interests.
For the study, the team used computer vision and artificial intelligence to examine the quality of photos on the social media site. Previous research conducted by a team from Penn Medicine in 2018 found that certain keywords picked up by artificial intelligence could predict depression three months before a person is diagnosed with it.
Detecting depression and anxiety is important as one study found it can lead to serious health issues like heart disease and arthritis. "Our findings are in line with a lot of other studies showing that psychological distress is not a strong predictor of many types of cancer. On top of highlighting that mental health matters for a whole host of medical illnesses, it is important that we promote these null findings. We need to stop attributing cancer diagnoses to histories of stress, depression and anxiety," one of the authors of the study Aoife O'Donovan from UC San Francisco, told a news portal.
Using algorithms, the team was also able to extract specific features like facial expressions and colours from more photos posted by more than 4,000 users.They also analysed each user's tweets and close to 887 tweeters were given a survey so researchers could score their levels of depression and anxiety. These users are also more likely to hide their negative emotions and suppress positive ones, according to the study's findings. Exhibiting a straight face in their profile pictures is one example. The results also revealed users with anxiety were more likely to have images with grayscale and low aesthetic cohesion. Meanwhile, depressed users were found to post photos only of themselves alone. These users also did not post photos that showed recreational activities or their interests.
For the study, the team used computer vision and artificial intelligence to examine the quality of photos on the social media site. Previous research conducted by a team from Penn Medicine in 2018 found that certain keywords picked up by artificial intelligence could predict depression three months before a person is diagnosed with it.
Detecting depression and anxiety is important as one study found it can lead to serious health issues like heart disease and arthritis. "Our findings are in line with a lot of other studies showing that psychological distress is not a strong predictor of many types of cancer. On top of highlighting that mental health matters for a whole host of medical illnesses, it is important that we promote these null findings. We need to stop attributing cancer diagnoses to histories of stress, depression and anxiety," one of the authors of the study Aoife O'Donovan from UC San Francisco, told a news portal.