A new study shows that analyzing the words a person uses can help predict worsening depression symptoms. Researchers used human evaluators and ChatGPT to analyze written responses. Both were able to accurately predict the severity of depression weeks later.
Traditional tools, like LIWC, did not perform as well as ChatGPT, which could understand the emotional tone based on word order and meaning. This discovery could help clinicians use AI to improve mental health evaluations, providing new ways to diagnose and predict mental health issues.
In the study, 467 participants answered open-ended questions and completed a depression survey (PHQ-9). Three weeks later, they completed the survey again. The researchers found that while LIWC could detect current depression severity, it couldn’t predict future symptoms. Human evaluators, however, could predict worsening symptoms by analyzing the participants’ language.
ChatGPT also performed well in this task, offering results similar to human evaluators. This success suggests that AI could complement psychological evaluations, helping clinicians better understand and predict mental health outcomes.
This research is part of an ongoing effort to explore the link between emotions and decision-making. The team hopes to extend this approach to other mental health conditions and longer study periods.
Researchers believe that AI tools like ChatGPT can be valuable in clinical settings, offering additional insights that help clinicians spend more time focusing on patient care.
Source: neurosciencenews
