What if there were a way to use data already at our disposal to detect children at risk for suicidal behavior? What if AI could help with that?
The alarming nationwide statistics on youth suicide speak for themselves. Data from the South Carolina Committee on Children and the U.S. Centers for Disease Control and Prevention report that suicide is the third leading cause of death among individuals ages 10 to 14 years old, and suicide rates for ages 10 to 24 rose 62% nationally from 2007 to 2021.
Unfortunately, South Carolina’s statistics align with those reported nationwide.
That challenge prompted MUSC to pursue a novel approach, with support from The Duke Endowment, that uses AI to develop predictive models capable of detecting complex patterns of suicide risk that traditional methods may miss. The models will analyze a range of information associated with suicide risk in youth, including known risk factors such as a history of depression, substance use, trauma exposure and social instability, as well as indicators such as suicidal ideation, mental health referrals, absenteeism and sleep disorders.
“Through this innovative approach, we are reimagining how we deliver early identification and intervention, with the ultimate goal of improving youth mental health outcomes,” said Jihad Obeid, M.D., who serves as principal investigator on the project. Obeid is a professor in the Department of Public Health Sciences and the assistant vice president for Research – AI Strategy and Research Intelligence at MUSC.
The multidisciplinary team that came together to create this system recognized several interconnected gaps in current prevention efforts. Despite that growing awareness, the ability to identify and intervene in at-risk youth had been severely hampered by fragmented data systems.
This lack of integrated data has prevented a comprehensive and holistic understanding of an individual’s risk factors. Consequently, warning signs are not recognized in time for interventions to be applied soon enough to prevent serious harm or suicidal behavior.
Bob Stevens, Ph.D., has collaborated with MUSC as a community engagement consultant on several research projects that affect South Carolina’s K-12 schools. His primary focus is on student behavioral health, and he regularly works with the S.C. State Department of Education, S.C. Office of Mental Health and individual school districts. He noted that this type of systematic data integration across schools and health systems was something he had never seen implemented before.
“The only time I have seen this was on a micro level, just with individual children. This program is a macro solution. This will provide for much more extensive analyses,” he said.
Joel Mensonides, executive director of system operations – Behavioral Health Network, was the first to approach Obeid and others on the team about the idea for this program.
“I am passionate about suicide prevention and was especially interested in how analytics could be used to help identify risk and support prevention efforts. I approached the team with the idea of using AI-enabled predictive analytics to advance our zero-suicide initiative. From there, we worked together to refine the project scope, focusing on youth and collaboration within the school system and MUSC’s existing school-based behavioral health services,” Mensonides said.
With assistance from South Carolina school districts and various healthcare systems, integrated data sources will come from electronic health records captured from a variety of medical organizations that do not communicate with one another; the Office of Mental Health’s collected data, including service utilization and clinical and outcomes data; and basic school information from the PowerSchool Student Information System, including attendance and demographics.
Data confidentiality is of the utmost importance to the team. They will protect student privacy by developing models without using direct identifiers, and all the data will be maintained in secure environments with protections comparable with those used for sensitive health data.
According to the South Carolina Behavioral Health Academy, schools are highly important in the collection of data because children are 21 times more likely to access mental health services in schools than in any other setting.
The three-year $1.23 million initiative will progress from partnership formation and secure data integration to predictive model development, evaluation and statewide dissemination.
In Year 1, the team will finalize school district and health system partnerships, obtain regulatory approvals, execute data-sharing agreements and establish secure data feeds and infrastructure.
In Year 2, the project will expand student enrollment, complete data integration and develop AI-based predictive models for suicide risk detection.
In Year 3, the team will evaluate and refine the models; share methods and findings with state partners; and disseminate results through workshops, publications, policy briefs and implementation guides.
A distinctive aspect of the project is its potential for long-term sustainability, with The Duke Endowment’s three-year investment helping to establish a model designed to continue well beyond the initial funding period.
LalithKumar Solai, M.D., is vice chair for clinical services within the Department of Psychiatry and Behavioral Sciences at MUSC. He also oversees MUSC’s school-based behavioral health services throughout South Carolina. In 2023, MUSC initiated behavioral health services in 23 schools across the state. “We are now in close to 150 schools with almost 20 therapists and a few child psychiatrists working within this service,” Solai said.
The team plans to work with community partners, such as federally qualified health centers that include HopeHealth in Florence, which will allow for a more comprehensive assessment of children who may face specific barriers that children in larger and more urban areas may not face.
We will focus on those individuals at elevated risk for suicide based on medical, behavioral and social indicators, including youth from varied racial and ethnic backgrounds, particularly those from underserved, low-income or rural communities who may face further barriers to accessing timely mental health care.
“We will focus on those individuals at elevated risk for suicide based on medical, behavioral and social indicators, including youth from varied racial and ethnic backgrounds, particularly those from underserved, low-income or rural communities who may face further barriers to accessing timely mental health care,” Solai said.
In addition to the project potentially lasting well into the future, the AI aspect made it stand out. However, Mensonides said the project’s focus on youth mental health is especially meaningful. “This effort is specifically designed to support young people, which makes it both timely and important. Our goal is to help address and prevent suicide within the youth population,” he said.
Solai noted that another aspect of the project’s most distinctive strengths is its multidisciplinary approach, connecting experts from areas such as data science, mental health, community-based youth support and others. Mensonides added that bringing together innovative leaders across these disciplines will be essential to the project’s success.
Stacy Warren, associate director of The Duke Endowment’s Health Care program, said they were thrilled to support this innovative effort. “This project builds on MUSC’s leadership in school-based youth mental health and has the potential to transform how we use data and AI to identify risks earlier, intervene sooner and improve outcomes for young people across the Carolinas,” she said.
“This issue is so critical,” concluded Stevens, the community engagement consultant. “We need to focus our attention on how to prevent youth suicide, and we need to do it now.”
Theresa Stratford
Theresa Stratford is the editor for MUSC’s newspaper, The Catalyst. She has worked as a writer and editor for 25 years since graduating from the University of South Carolina with a Bachelor of Arts degree in journalism and mass communications. As a writer, she has worked for the federal government in cybersecurity and for the Department of Energy. As both a writer and an editor, she has worked for a variety of syndications, including national magazines, such as Vogue Knitting, as well as many local Charleston and South Carolina publications.
Recent MUSC News stories
