According to a study published in the Journal of Anxiety Disorders by Pennsylvania State University researchers, artificial intelligence (AI) models can help doctors identify indicators that predict long-term recovery and better customize patient treatment.
Image Credit: everything possible/Shutterstock.com
Even after treatment, people with generalized anxiety disorder (GAD), a disorder marked by excessive worry every day for at least six months, have a significant relapse rate.
The researchers analyzed over 80 baseline parameters for 126 anonymized people with GAD diagnoses using a type of artificial intelligence called machine learning. These factors ranged from psychological and sociodemographic to health and lifestyle variables.
The information originated from the Midlife in the United States longitudinal study conducted by the U.S. National Institutes of Health. This project collects health data from continental US residents between the ages of 25 and 74 who were initially questioned in 1995 and 1996.
After nine years, the machine learning models found 11 characteristics that seem to be most crucial for predicting recovery and nonrecovery with an accuracy of up to 72%.
Prior research has shown a very high relapse rate in GAD, and there’s also limited accuracy in clinician judgment in predicting long-term outcomes. This research suggests that machine learning models show good accuracy, sensitivity and specificity in predicting who will and won’t recover from GAD. These predictors of recovery could be really important for helping to create evidence-based, personalized treatments for long-term recovery.
Candice Basterfield, Study Lead Author and Doctoral Candidate, Pennsylvania State University
The baseline variables were run through two machine learning models: a linear regression model, which examines the relationship between two variables and plots data points along a nearly straight line, and a nonlinear model, which branches out like a tree, splitting and adding new trees while plotting how it self-corrects prior errors.
The models revealed 11 variables important for predicting recovery or nonrecovery over a nine-year period, with the linear model beating the nonlinear one. The models also determined the relative importance of each variable in predicting recovery outcomes.
The researchers discovered that greater education level, older age, more friend support, higher waist-to-hip ratio, and higher positive affect, or feeling more cheerful, were the most relevant factors in recovery, in that order. Meanwhile, low mood, everyday discrimination, more sessions with a mental health professional in the previous 12 months, and more visits to medical professionals in the previous 12 months were the most important predictors of nonrecovery.
The researchers confirmed the model findings by comparing the machine learning predictions to the MIDUS data, and discovered that the expected recovery characteristics corresponded to the 95 persons who had no GAD symptoms at the end of the nine-year period.
The findings show that clinicians can utilize AI to discover these characteristics and tailor treatment for GAD patients, particularly those with multiple diagnoses, according to the authors.
According to Michelle Newman, senior author and psychology professor at Penn State, around 50% to 60% of patients with GAD suffer from comorbid depression. She noted that customized treatment could address both depression and anxiety.
Machine learning not only looks at the individual predictors but helps us understand both the weight of those predictors - how important they are to recovery or non-recovery - and the way those predictors interact with one another, which is beyond anything a human might be able to predict.
Michelle Newman, Study Senior Author and Professor, Pennsylvania State University
The study was unable to ascertain how long GAD lasted over the nine-year period, the researchers noted, because it is a chronic condition with periods of intense symptom manifestation that come and go. But, they said, the work sets the stage for more specialized treatments.
The researchers stated that the study could not assess the length of GAD over a nine-year period because it is a chronic disorder with times of severe symptoms that come and go. However, the study provides the framework for more personalized therapy, according to the authors.
Newman added, “This work helps us begin to understand more ways in which treatment could be personalized for specific individuals.”
The US National Institutes of Health, through the National Institute of Mental Health, financed this research.
Journal Reference:
Basterfield, C., and Newman, G., M., (2025) Development of a machine learning-based multivariable prediction model for the naturalistic course of generalized anxiety disorder. doi.org/10.1016/j.janxdis.2025.102978