Patient Perceptions of Artificial Intelligence in Diabetes Self-Management: Cross-Sectional Survey Study

Comprehensive Summary

Martini et. al investigates how people with diabetes perceive the role and usefulness of artificial intelligence (AI) in managing their condition while also looking into how AI fits into shared decision making and the relationship between a patient and provider. This cross-sectional study was conducted with 48 adults with diabetes in New Zealand in which participants rates seven tasks including data tracking, data interpretation, lifestyle management, and treatment decision making. Overall, AI appeared to be moderately useful especially for structured tasks. However, despite its perceived usefulness, the use of AI tools was still low and participants preferred healthcare providers for tasks requiring clinical judgement or personal reflection. Regression analysis also showed that perceived usefulness of AI predicted stronger preference for AI in several tasks, and if a relationship between a patient and provider is strong, there was a reduced preference for AI in treatment decisions and medication adherence. Overall, patients are open to AI when it supports data derived aspects of diabetes management, but they would still like to rely on human clinicians for nuanced or emotional tasks. This study suggests a triadic shared decision making model that integrates patients, providers, and AI as distinct contributors to diabetes management.

Outcomes and Implications

Patients may benefit from AI tools that help with tasks with data processing since these technologies can enhance monitoring and early detection of glycemic trends. Because patients prefer clinicians for contextualized or emotionally nuanced decisions, AI can be implemented as a supplement. Strengthening the relationship between patients and providers would also allow for less use of AI as well, making sure that medication adherence and treatment decisions are more personalized. Overall, patients, clinicians, and AI can work together to create a more personalized and efficient diabetes management if AI is implemented with attention to trust, transparency and patient autonomy.