Skip to main content

STORIES

New Study Using Machine Learning, Artificial Intelligence to Predict Teens’ Response to Depression Treatment

STORIES

New Study Using Machine Learning, Artificial Intelligence to Predict Teens’ Response to Depression Treatment

Anxiety and depression represent a growing healthcare concern for adolescents and young adults. Although antidepressant medication is part of the treatment plan for many suffering from depression, it can take weeks to see an effect and even longer to know if the treatment will work or if a different medicine should be tried. As a result, many patients and their families often suffer through weeks of trial and error before the right antidepressant is identified and shows effectiveness.

To address this problem, investigators at the Children’s Mercy Research Institute (CMRI) are collaborating with the Mayo Clinic on the GOLDILOKs PRISM study, which aims to match patients with an effective treatment as soon as possible and minimize the amount of time they may receive ineffective treatment.

In addition to working with researchers at the Mayo Clinic, the CMRI investigators are leveraging the expertise of the new Field-Based Physical Activity Measurement Core, Research Informatics, Precision Therapeutics, and clinicians and researchers in the Divisions of Adolescent Medicine and Developmental and Behavioral Sciences to design and conduct the GOLDILOKs PRISM study. Clinicians and staff in Adolescent Medicine and Teen Primary Care are working with the study team to help recruit patients and families.

Integrating the research component of this project into the usual clinical care patients receive sets this study apart from other depression studies in children and teens.

Catherine Koertje, MD
Clinical Research Project Manager, Clinical Pharmacology, Toxicology and Therapeutics

The goal of the GOLDILOKs PRISM study is to determine the accuracy of a machine learning/artificial intelligence-based model designed to predict a patient’s ultimate response to a selective serotonin reuptake inhibitor (specifically fluoxetine, or Prozac®) after only a few weeks of treatment. The study observes teens aged 12 to 18 diagnosed with depression over the first 3 months of their treatment with fluoxetine and then compares the observed treatment outcome with the outcome predicted by the machine learning/artificial intelligence-based model.

A secondary goal of this study is to further refine the model, developed by collaborators at the Mayo Clinic, and its ability to predict treatment outcomes as early as possible by identifying individual factors that may affect whether fluoxetine will improve depression symptoms. To identify these factors, the study team is collecting genetic, biochemical, proteomic, physiologic/lifestyle, and frequent symptom improvement data to generate “signatures” associated with clinical outcomes.

“Integrating the research component of this project into the usual clinical care patients receive sets this study apart from other depression studies in children and teens,” says Catherine Koertje, MD, Clinical Research Project Manager in the Department of Clinical Pharmacology, Toxicology and Therapeutics at Children’s Mercy. “Conducting the study in the same setting where the new knowledge generated by the study will be applied allows the benefits of the research to be realized within a shorter time frame than a typical clinical trial. Additionally, the study is designed to achieve an optimal balance between collecting large and diverse set of data from each participant over several weeks and minimizing the time and effort required from teens and their families to participate.”

Dr. Koertje’s CMRI colleagues on the GOLDILOKs PRISM study include Stephani Stancil, APRN, PhD, in the Division of Adolescent Medicine, and J. Steven Leeder, PharmD, PhD, Deputy Director of CMRI. The CMRI investigators are partnering with colleagues in the Field-Based Physical Activity Measurement Core at Children’s Mercy to establish and manage monitors worn by patients to measure their heart rate, sleep, and physical activity, and the team will then score and interpret data from those monitors. The Research Informatics team is downloading and storing physiologic and lifestyle data from the monitors and has developed custom reports to track monitor use daily.

In the future, the study team plans to work with colleagues in the Genomic Core Service Center at CMRI to analyze known genes and identify unknown genes related to the effects of selective serotonin reuptake inhibitors like fluoxetine on the body.

“There is no one-size-fits-all treatment for depression,” says Dr. Koertje. “The end product of this research initiative is to develop a tool that clinicians, patients, and their families can use to make more informed, individualized, and timely decisions about how to best treat a teen’s depression.”