Supporting Families of Children With Cancer Using a Virtual Assistant for Social Care: Development and Usability Study

Background Families of children with cancer face substantial financial, social, and emotional stressors, including unmet health-related social needs (HRSN) and toxic stress, which contribute to disparities. Virtual assistants (VAs) offer a scalable option to screen for HRSN and connect families to resources, but their design and acceptability in pediatric oncology are not well characterized. Objective The objective of this study was to evaluate a VA prototype designed to screen for HRSN and reduce toxic stress among families of children with cancer. Methods A 2-phase mixed methods study was conducted to co-design and evaluate a VA prototype for caregivers of children with cancer. In phase 1, a community advisory board (CAB) including health care providers and a parent advocate (n=8) iteratively refined VA content, tone, and interface through 3 virtual co-design sessions. In phase 2, caregivers (n=40) completed in-person VA prototype usability sessions that included structured observation, a semistructured interview, the System Usability Scale (SUS), a caregiver satisfaction survey, and a reaction card task. Social drivers of health, rurality, and neighborhood disadvantage were characterized using survey data and geocoded indices. Results CAB feedback refined prototype components including community resources, stress management, conversation history, and resource feedback. In phase 2, usability was high (SUS mean 82.6, SD 14.0); 95% (38/40) of caregivers rated the VA as easy to use and good to use, and 92.5% (37/40) reported that the VA motivated them to remain engaged during screening and resource identification. Qualitative analysis revealed that caregivers favored the VA’s simplicity and its potential to provide timely support, particularly shortly after diagnosis. Participants expressed strong preferences for electronic health record integration to streamline care, though they noted concerns regarding privacy and “app fatigue” as potential adoption barriers. Reaction card responses were predominantly positive (eg, “accessible,” “easy,” and “useful”). Greater socioeconomic vulnerability was associated with higher usability ratings, while rural and medically underserved caregivers reported lower confidence and more negative desirability attributes. Conclusions The stakeholder-informed VA to address HRSN and toxic stress in pediatric oncology was highly usable and acceptable in prototype testing. Future work should evaluate live deployment, resource linkage, toxic stress outcomes, and equity in larger, more diverse, and longitudinal studies.