AI, social media help track population-based mental health patterns

Researchers from Stony Brook University, Stanford University and the University of Pennsylvania have developed an AI-based system to track rates of depression and anxiety using data from social media posts, according to a recent study published in npj Digital Medicine.

The research team emphasized that depression and anxiety are major contributors to poor mental health for Americans. The ongoing mental health crisis is a significant driver of adverse outcomes like suicide and opioid-related deaths, spurring increased efforts to measure indicators of mental health.

The researchers further noted that traditional initiatives to assess mental health in populations entails using phone-based surveys asking individuals if they have experienced recent worry or sadness. However, these phone surveys are expensive and often do not yield enough information to reliably track mental health trends.

To address this, the research team set out to build an AI-driven system that can measure community-level mental health by incorporating language-based mental health assessments (LBMHAs) and social media language.

Using the model, the research team successfully analyzed almost one billion posts from more than two million X, formerly Twitter, users across 1,418 US counties. The system geo-located users, determined the language use patterns across posts and combined this information to estimate mental health trends.

The analysis revealed that the AI tool’s performance was highly accurate when matched to 2020 phone survey data.

“The main result of this study was a comparison of how well our AI model’s predictions lined up with survey-based methods and how computational methods enable new resolutions of mental health studies that were previously not possible,” explained lead author Siddharth Mangalik, a PhD student in computer science at Stony Brook, in a news release.

The system also outperformed phone-based survey methods by 10 percentage points when considering external factors such as income, education, housing and socialization.

Despite these initial successes, the research team noted that capturing mental health trends using language behaviors pulled from social media is not without its challenges.

“Social media measures allow us to track depression and anxiety – in principle – in real-time. Social media platforms are constantly changing in their leadership, policies, and how researchers can access data for the common social good,” said Johannes C. Eichstaedt, PhD, assistant professor in Psychology and Human-Centered AI at Stanford.

However, the researchers indicated that speech and language patterns present valuable insights into a person’s emotional state.

“This study provides a new tool to describe and understand public mental health in a way that was unimaginable just five years ago,” stated co-author Sean Clouston, PhD, a professor in the Program in Public Health and in the Department of Family, Population, and Preventive Medicine in the Renaissance School of Medicine (RSOM) at Stony Brook University. “We hope it can soon be used by clinicians, mental health providers and others to help improve public mental health in the future.”

Moving forward, the research team will continue to evaluate the AI’s performance as language patterns and social media platforms change in the coming years. The team also recommended that public health stakeholders consider using both survey data and language-based assessments to measure community mental health, as doing so could help combat participants’ tendency to under-report stigmatized traits, like symptoms of mental health conditions.