Exploring the Perspectives of Pediatric Health Care Providers, Youth Patients, and Caregivers on Machine Learning Suicide Risk Classification: Mixed Methods Study

Editorial reading guide · Respect

How would our family be involved if an algorithm flagged a young person’s risk?

Participants saw opportunities for earlier conversations but raised privacy, consent, coercive-care, and misclassification concerns. These values affect whether a risk flag supports a young person or damages future engagement with care.

Study context

Forty-five pediatric providers answered at least one survey question, eight were interviewed, and patient-side interviews included nine patients and one caregiver.

Age / stage: Patient interview ages 8–25 years; limited representation under eighteen

Read with three questions

1. Who wanted access to risk information, and under what safeguards?
2. How could coercive responses undermine future care engagement?
3. How does the small number of caregivers and minors limit claims about family preferences?

Keep the limits in view

The patient-side sample spanned childhood to young adulthood and included only one caregiver, with few minors. One institutional setting, survey attrition, and framework-based interviews limit generalization; the study did not test prediction accuracy or harm reduction.