Researchers at the Children's Hospital of Eastern Ontario (CHEO) are harnessing artificial intelligence to address a critical challenge: the high rate of children and youth who return to the emergency department within six months of an initial mental health visit. This innovative approach aims to reduce emergency room revisits by identifying patients at highest risk and ensuring they receive more comprehensive support.
High Revisit Rates Pose Significant Risks
Hospital officials report that CHEO has a higher rate of emergency visits for mental health needs compared to other similar pediatric institutions across Canada. According to Dr. Kathleen Pajer, director of the CHEO Research Institute's Precision Child and Youth Mental Health Collaboratory, approximately 20 to 30 percent of these young patients return to emergency for urgent help, sometimes repeatedly.
These revisits are particularly concerning because children and youth mental health patients who make multiple emergency trips within six months face higher risks for adverse outcomes, including increased rates of suicide, self-harm, and overall mortality. Additionally, they contribute to the growing demand for care in already strained emergency departments.
Rising Mental Health Visits and the Pandemic Impact
Child and youth mental health visits to emergency departments grew significantly from the mid-2000s until the COVID-19 pandemic, increasing between two and five times. Although rates began to decline between 2020 and 2022, mental health visits did not drop as much as other emergency visits, highlighting the persistent need for specialized care.
Dr. Pajer's collaboratory integrates biological, social, environmental, and clinical factors to create a more complete understanding of each young person's mental health needs, providing a foundation for the AI research.
Machine Learning Outperforms Traditional Methods
Recent research published in BMC Medical Informatics and Decision Making details the development of a predictive algorithm using machine learning, a type of artificial intelligence. The study found that this AI system performed better than doctors and other clinicians at predicting which patients were most likely to return to the emergency department.
The computer algorithm achieved just over 70 percent accuracy in identifying individuals who would revisit the emergency room by analyzing specific patient characteristics and generating predictions based on them.
Refining the Algorithm for Clinical Use
Dr. Pajer emphasized that there is still work to be done to refine the results. Researchers plan to incorporate more variables and clinical input to enhance the algorithm's accuracy. Once finalized, clinicians at CHEO will help develop a protocol for implementation when a young patient is identified as high risk.
This protocol may include enhanced support measures such as improved connections to outpatient programs or more frequent follow-up contacts to assess whether patients are receiving the necessary help. The ultimate goal is to prevent emergency revisits by providing timely, targeted interventions.
Addressing a Growing Healthcare Challenge
The integration of artificial intelligence into pediatric mental health care represents a promising step toward addressing the swelling demand for emergency services. By leveraging technology to predict and prevent revisits, CHEO aims to improve outcomes for vulnerable young patients while alleviating pressure on emergency departments.
This research underscores the potential of AI to transform healthcare delivery, particularly in areas where traditional approaches have struggled to keep pace with growing needs. As the algorithm is refined and implemented, it could serve as a model for other institutions facing similar challenges in pediatric mental health care.
