Understanding Adolescents’ Perceptions of Benefits and Risks in Health AI Technologies through Design Fiction

Research Background and Issues

• What problems or challenges have the authors identified?

  • The understanding of adolescents' perceptions of health artificial intelligence (Health AI) has not been thoroughly explored.
  • While the potential of Health AI in medical care and personal health management has garnered attention, concerns remain regarding privacy risks, trust issues, and its applicability to specific groups, such as adolescents.
  • Adolescents are in a transitional phase of developing unique health behaviors, necessitating the design of Health AI tailored to their specific needs.

• Why is this issue important?

  • Adolescents are at a critical stage for establishing long-term health behaviors, gaining autonomy in personal health management, and often relying on technology (e.g., the internet and social media) for health information.
  • As Health AI becomes more prevalent, it is likely to have a profound impact on adolescents' health management practices.

• Research Motivation and Related Work

  • Previous studies have primarily focused on adult users (e.g., clinicians, patients, and parents) or patients with specific health conditions, neglecting adolescents' perspectives on Health AI.
  • This study aims to fill this research gap by exploring adolescents' views on Health AI technologies and providing guidance for future design.

Solutions

• What methods or solutions did the authors propose?

  • The authors employed the Design Fiction method, using four fictional design scenarios to explore the potential benefits and risks of Health AI technologies for adolescents.
  • These fictional scenarios were used to guide adolescents in imagining future use cases of Health AI, thereby eliciting their genuine feedback.

• What is innovative about this solution?

  • The use of the Design Fiction method allows for insights into actual needs and potential impacts before the technology is fully implemented.
  • The study incorporates dual perspectives from clinical and personal health domains, offering broader coverage.
  • It conducts a detailed investigation into adolescents' unique needs, such as privacy and learning health management skills.

• What are the implementation steps? What key techniques were used?

  • Scenario Development: Collaborated with pediatricians and combined literature and clinical practices to design four scenarios covering current and future use cases.
  • Participant Recruitment: Recruited 16 adolescents for interviews through summer programs at two high schools in the western United States.
  • Semi-Structured Interviews: Presented each fictional scenario via Zoom and collected feedback.
  • Data Analysis: Conducted thematic analysis using open coding to extract adolescents' attitudes, evaluations of pros and cons, and suggestions.

Research Findings

• What specific findings were obtained?

  • Adolescents expressed cautious optimism about Health AI, identifying the following benefits:
    • Providing more efficient health management tools.
    • Assisting doctors in improving diagnostic and treatment efficiency.
    • Supporting personalized health goals (e.g., exercise, nutrition management).
  • Adolescents also voiced concerns about certain risks:
    • Privacy issues, including the potential for information to be leaked to the public or parents.
    • Low predictive accuracy of AI in mental health and critical care domains.
    • Reservations about decisions made solely by AI (e.g., medication recommendations).

• What advantages does it have compared to existing solutions?

  • This study is the first to specifically explore adolescents' perceptions of Health AI, ensuring that this group's voice is considered in technology design.
  • It provides comprehensive insights into adolescents' specific needs and perspectives (e.g., learning and privacy needs), enabling more targeted future designs.

• What were the experimental or evaluation results?

  • Adolescents emphasized the need for human involvement in all Health AI functions, including providing technical feedback, emotional support, and decision-making oversight.
  • Their varying sensitivity to different types of data privacy (e.g., medical records versus body recordings) revealed the contextual dimensions of privacy risks.

• Limitations and Future Directions

  • Limitations:
    • The sample was skewed toward older adolescents (average age 16) and females, which may not fully represent the broader adolescent population.
    • Participants did not have severe health conditions, potentially limiting feedback from adolescents with significant health issues.
  • Future Directions:
    • Investigate the perceptions of Health AI among adolescents of different age groups or those with chronic illnesses.
    • Explore how race and cultural backgrounds influence perceptions of AI.
    • Promote racial inclusivity and data transparency in Health AI design to mitigate risks and enhance applicability.