EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

Optimizing Human–AI Interaction for Older Adults: A User Interface Experience Study on Accessible Medical Diagnostic Interfaces

Wenfang Pan, Chuifeng Dong

IEEE Access · 2026 · 14 · 122664-122676

DOI: 10.1109/access.2026.3718379 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Access, 2026. Verfasst von Wenfang Pan und 1 weiteren Autor:innen. Die Arbeit nennt 31 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This paper reports an HCI study of an accessible Human–AI diagnostic-information review interface for older adults. The study is motivated by user needs associated with cognitive aging, reduced visual acuity, motor-control variability, heterogeneous digital literacy, and WCAG-informed accessibility requirements. Rather than presenting a new autonomous diagnostic model, federated-learning algorithm, or systems/ML contribution, the manuscript examines how these older-adult accessibility requirements can be translated into a participant-facing interface prototype and evaluated through usability, workload, task-efficiency, and interaction-error measures. The implemented prototype uses three supporting components: (A) interface-component identification and diagnostic-indicator prioritization to reduce visual clutter; (B) rule-constrained presentation adaptation to adjust visual hierarchy, contrast, layout density, and confirmation prompts; and (C) a simulation-based, privacy-motivated feedback benchmark for studying whether anonymized interaction summaries can support distributed interface refinement. The reviewed health information consisted of routine home-monitoring data for three common chronic-condition domains familiar to older adults—blood-pressure (hypertension) trends, blood-glucose (diabetes) readings, and heart-rate and medication-adherence reminders—rather than oncology screening or image-based diagnostic results. A within-subjects crossover pilot study with older adults (N=42, aged 65–82) compared the proposed adaptive interface with a static non-adaptive diagnostic-information interface using the same health-information content, task templates, apparatus, and participant protocol. Participant-level HCI outcomes included System Usability Scale (SUS), NASA Task Load Index (NASA-TLX), task-completion time, error rate, gaze fixation, and interaction burden. Results are reported as preliminary within-subjects HCI evidence rather than as definitive population-level effectiveness evidence. The proposed interface showed higher perceived usability than the static baseline (SUS: 86.4 vs. 62.1; paired mean difference = 24.3, 95% CI $[{20.6, 28.0}]$ , $t(41)=13.12$ , $p\lt 0.001$ , Cohen’s $d_{z}=2.03$ ) and lower perceived workload on NASA-TLX (relative reduction = 41%; paired mean difference = 25.4, 95% CI $[{21.8, 28.9}]$ , Wilcoxon $W=903$ , $p\lt 0.001$ , rank-biserial $r_{rb}=1.00$ ) in the recruited cohort.

Zitieren

APA

Pan, W., & Dong, C. (2026). Optimizing Human–AI Interaction for Older Adults: A User Interface Experience Study on Accessible Medical Diagnostic Interfaces. IEEE Access, 14, 122664-122676. https://doi.org/10.1109/access.2026.3718379

Vancouver

Pan W, Dong C. Optimizing Human–AI Interaction for Older Adults: A User Interface Experience Study on Accessible Medical Diagnostic Interfaces. IEEE Access. 2026;14:122664-122676. doi:10.1109/access.2026.3718379

BibTeX
@article{Pan2026,
  author = {Pan, Wenfang and Dong, Chuifeng},
  title = {Optimizing Human–AI Interaction for Older Adults: A User Interface Experience Study on Accessible Medical Diagnostic Interfaces},
  journal = {IEEE Access},
  year = {2026},
  volume = {14},
  pages = {122664--122676},
  doi = {10.1109/access.2026.3718379}
}
Kostenlos starten →