Sao Paulo: Work is underway to create anxiety and depression prediction models, using artificial intelligence (AI) and Twitter, one of the largest in the world social media platforms, which could detect signs of these diseases earlier clinical diagnosis according to the researchers.
Researchers of the University of Sao Paulo (USP) in Brazil said that preliminary findings from the model suggested the possibility of detecting the probability that a person will develop depression based solely on your social media friends and followers.
The findings are published in the journal Language resources and assessment. While there are multiple studies involving natural language processing (NLP) focused on depression, anxiety and bipolar disorder, most of these analyzed texts in English and did not match the profiles of Brazilians, the researchers said.
The first step of this study involved the construction of a database, called SetembroBR, of information related to a corpus of 47 million publicly published texts in Portuguese and the network of connections among 3,900 Twitter users. These users had reportedly been diagnosed or treated for mental health issues prior to the survey. The tweets were collected during the COVID-19 pandemic.
“First, we compiled timelines manually, analyzing the tweets of about 19,000 users, the equivalent of the population of a town or small city.
“We then used two data sets, one for users who reported being diagnosed with a mental health problem and one randomly selected for control purposes. We wanted to distinguish between people with depression and the general population,” said Ivandre Paraboni, last author of the study. article and professor at USP.
Because people with mental health problems tended to follow certain accounts, such as discussion boards, influencers and celebrities who publicly acknowledge their depression, the study also collected tweets from friends and followers.
The second step, still ongoing, has provided some preliminary findings, such as the possibility of detecting the probability that a person will develop depression based solely on their friends and followers on social networks, without taking their own posts into account.
Following pre-processing of the corpus to maintain the original texts by removing non-standard characters, the researchers implemented deep learning (AI) to create four text classifiers and word embeddings (context-dependent mathematical representations of relationships between words) using models based on bidirectional encoder representations of transformers (BERT), a machine learning algorithm employed for NLP.
These models correspond to a neural network which learns context and meaning by monitoring relationships of sequential data, such as words in a sentence. The training input consisted of a sample of 200 randomly selected tweets from each user.
The researchers found that among the models, BERT performed best in terms of predicting depression and anxiety. They said that because the models analyzed sequences of words and whole sentences, it was possible to observe that people with depression, for example, tended to write about topics related to themselves, using first-person verbs and phrases, as well as topics such as how death, crisis and psychology.
“The signs of depression that can be picked up during a doctor’s visit are not necessarily the same as those that show up on social media,” Paraboni said.
“For example, the use of the first person singular pronouns I and me was very evident, and in psychology this is considered a classic sign of depression. We also observed the frequent use of the heart emoji by depressed users.
“This is widely felt to be a symbol of affection and love, but perhaps psychologists haven’t characterized it as such yet,” Paraboni said.
The researchers are now expanding the database, refining their computational techniques, and updating the models to see if they can produce a tool for future use in screening potential patients with mental health problems and helping families and friends of young people in risk of depression and anxiety.