Related Experiment Video
Updated: Sep 10, 2025
Implementation of a Real-Time Psychosis Risk Detection and Alerting System Based on Electronic Health Records using CogStack
Published on: May 15, 2020
Exploring the Perspectives of Pediatric Health Care Providers, Youth Patients, and Caregivers on Machine Learning
Rohan R Dayal1, Pua Lani Yang1, Laura Nicole Sisson2
1Department of International Health, Bloomberg School of Public Health, Johns Hopkins University, 615 N Wolfe St, Baltimore, MD, 21205, United States, 1 9172837220.
Machine learning (ML) models show promise for identifying youth suicide risk, but implementation requires careful consideration of provider, patient, and caregiver perspectives. Addressing concerns about confidentiality and accuracy is key for successful integration into clinical practice.
Area of Science:
- Adolescent Health
- Mental Health Technology
- Clinical Informatics
Background:
- Youth suicide is a critical public health issue, ranking as the second leading cause of death for ages 10-24 in 2023.
- Machine learning (ML) applied to electronic health records (EHRs) offers potential for improved suicide risk identification.
- Further research is needed to understand stakeholder perspectives for effective ML tool implementation.
Purpose of the Study:
- To explore patient, caregiver, and pediatric healthcare provider perspectives on ML-based suicide risk models.
- To inform the design and implementation of a suicide risk model and clinical workflow for at-risk youth.
- To enhance the quality of care for youth experiencing suicidal ideation.
Main Methods:
- Convergent mixed methods study evaluating pediatric provider perspectives via surveys and interviews.
- Sequential mixed methods study exploring youth patient and caregiver perspectives through qualitative interviews.
- Data analyzed using descriptive statistics, template analysis, and joint display for integration.
Main Results:
- Providers, patients, and caregivers expressed interest in ML for suicide risk identification.
- ML tools could improve care by addressing limitations in manual screening and inter-team communication.
- Concerns include increased mental health demand, confidentiality, accuracy, liability, data safety, and patient autonomy.
Conclusions:
- ML-based suicide risk models have conditional acceptability among stakeholders.
- Successful implementation necessitates integrating provider insights for trust and clinician empowerment.
- Upholding patient and caregiver perspectives is crucial for meeting needs and ensuring positive engagement.
