Presentation
AI for UI: Designing Error-Tolerant Interfaces for Home Dialysis using Artificial Intelligence
SessionPoster Session 1
DescriptionHome healthcare technologies, especially those supporting high-risk, self-administered treatments such as at-home dialysis, present unique challenges for human factors professionals. Patients and caregivers must perform complex procedures in domestic environments with minimal supervision, often while managing anxiety, fatigue, comorbidities, and competing daily demands. Errors in setup, sequencing, or sterile technique can have severe consequences, yet many existing devices and training protocols offer limited support for anticipating, preventing, or mitigating these risks. Prior research applying UX heuristics to home dialysis has revealed recurring vulnerabilities in training quality, error tolerance, and interface design.
Building on that foundation, this research leverages artificial intelligence (AI) to systematically generate, classify, and curate design improvements that support error-tolerant user interface solutions for at-home dialysis systems. The proposed work bridges “AI to UI” by using AI as an accelerator to utilize insights from device manuals, regulatory databases, incident reports, and prior usability studies as actionable, reusable UI patterns. This pattern library is not intended to replace human-centered evaluation, but rather to provide a practical toolkit that HF experts, device designers, and clinicians can draw upon when developing or refining home dialysis technologies.
The importance of this research lies in its dual focus on immediacy and scalability. In the near term, the proposed design mitigations offer concrete interface improvements that can reduce error likelihood and strengthen patient autonomy in home dialysis. Long-term, the methodology establishes a replicable generative-AI pipeline that can produce similar heuristic-driven design libraries for a wide range of home healthcare contexts, such as infusion pumps and remote monitoring platforms. This approach enhances design consistency, accelerates the translation of human factors principles into real-world products, and underscores the role of HF professionals in guiding safe, responsible AI use in healthcare.
Three key takeaways emerged from our study: AI is not a replacement for human factors judgment, but an effective support tool in safety-critical environments where usability directly affects outcomes. Error-tolerant UI design can be codified into reusable patterns, improving consistency, reducing cross-vendor variability, and strengthening patient-facing device reliability. HF-driven, preclinical research can meaningfully advance safety and usability even before patient trials occur, enabling rapid, ethical, and evidence-based design iteration.
This study extends prior findings in dialysis usability and addresses unmet needs in training, privacy, and error tolerance by demonstrating how AI can mine and classify design opportunities across diverse data sources. Concrete examples of proposed UI patterns, including their rationale, anticipated benefits, and limitations, are presented. Broader implications for healthcare human factors include the value of developing design-mitigation libraries as companion resources to standards and regulatory guidance.
The broader impact of this work is its potential to establish design mitigation libraries as a new class of deliverable within healthcare human factors. Just as safety checklists standardized clinical best practices, AI-curated and expert-validated pattern libraries can help ensure safer, more intuitive interfaces across the expanding landscape of home healthcare. Our findings reinforce the role of human factors in driving technological improvements that empower patients, protect safety, and build trust in an era of rapidly expanding AI use.
Building on that foundation, this research leverages artificial intelligence (AI) to systematically generate, classify, and curate design improvements that support error-tolerant user interface solutions for at-home dialysis systems. The proposed work bridges “AI to UI” by using AI as an accelerator to utilize insights from device manuals, regulatory databases, incident reports, and prior usability studies as actionable, reusable UI patterns. This pattern library is not intended to replace human-centered evaluation, but rather to provide a practical toolkit that HF experts, device designers, and clinicians can draw upon when developing or refining home dialysis technologies.
The importance of this research lies in its dual focus on immediacy and scalability. In the near term, the proposed design mitigations offer concrete interface improvements that can reduce error likelihood and strengthen patient autonomy in home dialysis. Long-term, the methodology establishes a replicable generative-AI pipeline that can produce similar heuristic-driven design libraries for a wide range of home healthcare contexts, such as infusion pumps and remote monitoring platforms. This approach enhances design consistency, accelerates the translation of human factors principles into real-world products, and underscores the role of HF professionals in guiding safe, responsible AI use in healthcare.
Three key takeaways emerged from our study: AI is not a replacement for human factors judgment, but an effective support tool in safety-critical environments where usability directly affects outcomes. Error-tolerant UI design can be codified into reusable patterns, improving consistency, reducing cross-vendor variability, and strengthening patient-facing device reliability. HF-driven, preclinical research can meaningfully advance safety and usability even before patient trials occur, enabling rapid, ethical, and evidence-based design iteration.
This study extends prior findings in dialysis usability and addresses unmet needs in training, privacy, and error tolerance by demonstrating how AI can mine and classify design opportunities across diverse data sources. Concrete examples of proposed UI patterns, including their rationale, anticipated benefits, and limitations, are presented. Broader implications for healthcare human factors include the value of developing design-mitigation libraries as companion resources to standards and regulatory guidance.
The broader impact of this work is its potential to establish design mitigation libraries as a new class of deliverable within healthcare human factors. Just as safety checklists standardized clinical best practices, AI-curated and expert-validated pattern libraries can help ensure safer, more intuitive interfaces across the expanding landscape of home healthcare. Our findings reinforce the role of human factors in driving technological improvements that empower patients, protect safety, and build trust in an era of rapidly expanding AI use.
Event Type
Poster Presentation
TimeMonday, March 234:45pm - 6:15pm EDT
LocationRhinelander Gallery
Medical and Drug Delivery Devices