 |
If replicated, these results could help clinicians identify patients most at risk for poor outcomes and initiate treatment to prevent it. “[T]hese predictive models could inform the personalized prevention of functional impairment in patients with clinical high-risk states and patients with recent-onset depression,” wrote Nikolaus Koutsouleris, M.D., of Ludwig-Maximillian University in Germany and colleagues.
The researchers analyzed data on 116 individuals considered to be at high risk for psychosis and 120 patients with recent-onset depression using machine learning. Machine learning is a new technology that uses computer programs to analyze extremely large amounts of data and develop models that can predict certain kinds of outcomes for individual patients.
Using data collected at baseline, the machine learning models were able to successfully predict one-year social functioning in 76.9% of patients in high-risk states and 66.2% of patients with recent-onset depression. When combined with certain brain imaging findings, machine learning models successfully predicted social outcomes at one year in up to 83% of patients in high-risk states and 70% of patients with recent-onset depression.
The computer models did a better job of predicting social functioning outcome than did individual clinicians. “We observed that combined models integrating clinical and brain structural data outperformed human clinical raters, suggesting that these models could improve the prognostic process beyond the current level,” Koutsouleris and colleagues wrote.
For related information, see the Psychiatric News article “Early Social Functioning May Predict Long-Term Outcome in Psychosis.” For more about machine learning, look for the next edition of Psychiatric News.
(Image: iStock/FatCamera)
