A team of researchers at Oregon Health & Science University were awarded a landmark five-year, $16.4 million grant from the National Institute of Mental Health to develop and test data-driven approaches that can more precisely predict mental health diagnoses and outcomes in children.
A team of researchers at Oregon Health & Science University were awarded a landmark five-year, $16.4 million grant from the National Institute of Mental Health to develop and test data-driven approaches that can more precisely predict mental health diagnoses and outcomes in children.
For over a decade, experts at the OHSU Center for Mental Health Innovation have used machine learning to develop advanced computational models that can improve clinical prediction of a variety of mental health conditions across childhood and adolescence, including ADHD, anxiety, depression and substance use disorder.
These models — which provide insight on the predicted age of condition onset, severity and prognosis, ideal treatments and more — can serve as a valuable tool in clinical decision-making processes, and may ultimately inform more effective mental health intervention and prevention efforts, said Bonnie Nagel, Ph.D., professor of psychiatry in the OHSU School of Medicine and co-director of the Center for Mental Health Innovation.
After demonstrating the models’ success in the lab, the grant takes the work of OHSU scientists to the next level by allowing the team to transition its research algorithms to a clinical setting and determine the real-world effects they may have on clinical decisions and mental health outcomes.
Youth mental health crisis
Half of the world’s population will experience a mental health issue at some point in their lives, a 2023 study found, making mental health the largest source of disability in the world. The crisis is most severe for youth, who are experiencing unprecedented rises in mental health problems. In Oregon alone, about 50,000 children are living with a mental health condition and many face extremely poor access to care.
The life challenges that accompany a mental health condition persist and often multiply with development, leading to a variety of poor long-term outcomes such as incarceration, unemployment, substance use, and early death due to suicide, accident or physical health complications, Nagel said.
The health care system has struggled to effectively identify those at risk of a mental health condition and prevent poor outcomes among children seen in pediatric or psychiatry clinics. To address the severity and urgency of this crisis, OHSU researchers say it’s critical to identify new approaches in clinical care, including bringing data science to patient assessment and care delivery.
“The reality is that our children are suffering, and their suffering has worsened over the past decade. We need to think outside of the box and challenge the systems and approaches we’ve previously relied on,” said Joel Nigg, Ph.D., professor of psychiatry in the OHSU School of Medicine and co-director of the Center for Mental Health Innovation. “Twenty years ago, we didn’t have the computer power or scientific knowledge about psychopathology that we do now. This work is a very exciting step forward that harnesses the past two decades of scientific progress.
“It will take time to see the effects of these data-driven approaches and novel clinical measures,” he continued, “but they have the potential to create significant impacts, including lowering health care costs, improving care efficiency and access, and most importantly, saving lives.”
Most mental health conditions emerge during childhood and adolescence, making early detection and treatment key to improving outcomes. Through this grant, the National Institute of Mental Health, part of the National Institutes of Health, will fund additional sites to investigate data-driven approaches to mental health for various populations; OHSU is one of only a few sites that will focus specifically on children and adolescents.
Leveraging the power of data
During the initial stages of the five-year grant, researchers will optimize prediction models in the lab by leveraging large national research and medical databases and OHSU-developed research cohorts. These models comprise the observable traits of a patient, including temperament; cognitive measures like working memory and executive function; mental health characteristics like inattention; and environmental factors like proximity to health care or pollution exposure risk.
The model predictors will then be collected in clinics at collaborating sites, with a goal of enrolling 7,500 patients ages 7 through 17. To ensure data are unbiased and account for health care disparities, OHSU will work with hospitals across the country to collect patient data representing different socio-demographic, ethnic and racial backgrounds.
Collaborating sites include Seattle Children’s Hospital, Massachusetts General Hospital and Indiana University. Additional partners, including State University of New York and Purdue University, will be involved in data analytic efforts.
Finally, using the data collected across clinics, the research team will analyze and evaluate how often high-confidence predictions can be made, how often additional assessment is needed, and the added value and overall accuracy of the models. If successful, the proposed work would prove the readiness of these models for clinical use and create an actionable, cost-effective approach for health systems to adopt.
“We consider this a formative grant in achieving the mission of the center, which is to advance scientific discoveries in mental health and rapidly translate those into real-world practices,” Nagel said. “It’s tremendously exciting to be taking that next step of bringing together science and clinical care. It represents a new frontier in how we approach mental health.”
Research reported in this publication was supported by the National Institutes of Health under Award Number U01MH135970. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
