文献速递烧伤创面修复

人工智能增强多光谱成像烧伤评估工具的多中心可用性启发式评价

Usability of an Artificial Intelligence-Enhanced Multispectral Imaging Tool for Burn Assessment: A Multicenter Retrospective Heuristic Evaluation.
PubMed 文献 · J Burn Care Res · 2026-01-01

各领域均分3.93–4.35、中位数≥4.0、正面反应65.3%–87.8%,职业与中心间无显著差异,支持常规使用。

要点

  • AI-MSI在烧伤评估中可用性高,各领域均分3.93–4.35、中位数≥4.0。
  • 医生与护士、各中心间可用性无显著差异,正面反应65.3%–87.8%。
  • 自由文本建议优化屏幕提示、减轻成像头重量、延长电池及简化数据录入。

核心结论

  • AI-MSI在烧伤评估中跨角色、跨中心显示高可用性,支持在门诊、病房和手术室常规使用。

研究设计与样本

  • 采用多中心改良Nielsen启发式框架,12项Likert问卷映射6个领域并附自由文本。

主要发现

  • 各领域均分3.93–4.35,中位数≥4.0,正面反应(评分≥4)比例为65.3%–87.8%。
  • 职业间与中心间无显著差异;效应量小至中等,Match & Standards效应最大(ε2=0.15),校正后无两两差异。
  • 自由文本突出速度、输出清晰度、便携性和工作流程契合,并建议更清晰的屏幕提示、更轻的成像头、更长电池寿命和简化数据录入。

关键数值与推荐

  • 正面反应定义为评分≥4;各领域均分3.93–4.35,中位数≥4.0,正面反应65.3%–87.8%。
  • 统计显著性设定为P<.05,使用JASP 0.95.1分析;Mann-Whitney U检验比较职业,Kruskal-Wallis检验中心变异,必要时Dunn两两检验加Bonferroni校正。

临床怎么用

  • 可将AI-MSI纳入门诊、病房和手术室的常规烧伤评估流程。
  • 依据可用性结果设计培训与临床路径,并针对软件提示和硬件(成像头重量、电池)进行优化。

局限性

  • 研究为回顾性设计,证据限于可用性评价,未报告深度评估准确性等临床结局。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

Accurate burn-depth assessment remains difficult. Artificial intelligence-enhanced multispectral imaging (AI-MSI) offers objective, image-based predictions, but evidence on day-to-day usability is limited. The aim of this study is to evaluate the usability of an AI-MSI device in routine practice using a multicenter, modified Nielsen heuristic framework. A retrospective, mixed-methods study was conducted across 5 UK burn services (September 2024 to August 2025). Doctors and nurses who used the device completed a 12-item Likert survey mapped to 6 domains and free-text. Outcomes include domain means ± SD, medians (IQR), and proportions of positive responses (scores ≥ 4). The Mann-Whitney U test was used to compare professions; Kruskal-Wallis tested site variability; and Dunn pairwise tests with Bonferroni correction were used where indicated. JASP 0.95.1 was used for analysis with statistical significance set at P < .05. Forty-nine burn-team members (21 doctors, 28 nurses) participated. Domain means ranged from 3.93 to 4.35; medians were ≥ 4.0; and positive responses 65.3%-87.8%. There were no significant differences between professions or centers. Effect sizes were small to moderate; Match & Standards showed the largest effect (ε2 = 0.15) with no pairwise differences after correction. Free-text highlighted speed, clarity of outputs, portability, and workflow fit, and suggested clearer on-screen cues, a lighter imaging head, longer battery life, and streamlined data entry. Artificial intelligence-enhanced multispectral imaging showed high usability across roles and centers, supporting routine use in clinics, wards, and theaters. As the first multicenter heuristic usability evaluation of AI-MSI in burns, the study provides implementation-focused evidence to guide training, pathway design, and targeted software and hardware refinements.

来源:J Burn Care Res,查看原文。版权归原作者所有。

引用

  1. Abdulla H, Tan P, Bache SE, et al. Usability of an Artificial Intelligence-Enhanced Multispectral Imaging Tool for Burn Assessment: A Multicenter Retrospective Heuristic Evaluation. J Burn Care Res. 2026. PMID: 42252144. doi:10.1093/jbcr/irag087

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。