AI for UI: Designing Error-tolerant Interfaces for Home Dialysis using Artificial Intelligence
TL;DR
GenAI-assisted processes can provide rapid, actionable design mitigations that reduce error likelihood and enhance patient autonomy, establishing a replicable pipeline for producing heuristic-driven design libraries across diverse medical device contexts.
Abstract
Home healthcare technologies, particularly those supporting high-risk, self-administered treatments like at-home dialysis, present significant challenges for human factors professionals. Patients must navigate complex procedures while managing anxiety, fatigue, and comorbidities. Errors in process sequence or use can have severe consequences. This research leverages generative artificial intelligence to systematically generate, classify, and curate design patterns that support error-tolerant user interface solutions. Using a specialized Gemini Gem implementation, 70 design principles grounded in IEC 62366-1 and FDA guidance were codified, alongside a library of 50 UX design patterns addressing critical HCI concerns such as trust calibration, cognitive load, and alarm confusion. The GenAI-driven evaluation demonstrated a high degree of accuracy in assessing compliance across four existing at-home dialysis interfaces. The results indicate that GenAI-assisted processes can provide rapid, actionable design mitigations that reduce error likelihood and enhance patient autonomy. Long-term, this methodology establishes a replicable pipeline for producing heuristic-driven design libraries across diverse medical device contexts, supporting HF professionals in guiding safe, scalable, and responsible AI use in healthcare.
