The Smith Institute for Urology, The North Shore, Long Island Jewish Health System, New Hyde Park, NY 11042, USA.
Ann Surg. 2010 Jul;252(1):177-82. doi: 10.1097/SLA.0b013e3181e464fb.
Currently, surgical skills assessment relies almost exclusively on subjective measures, which are susceptible to multiple biases. We investigate the use of eye metrics as an objective tool for assessment of surgical skill.
Eye tracking has helped elucidate relationships between eye movements, visual attention, and insight, all of which are employed during complex task performance (Kowler and Martins, Science. 1982;215:997-999; Tanenhaus et al, Science. 1995;268:1632-1634; Thomas and Lleras, Psychon Bull Rev. 2007;14:663-668; Thomas and Lleras, Cognition. 2009;111:168-174; Schriver et al, Hum Factors. 2008;50:864-878; Kahneman, Attention and Effort. 1973). Discovery of associations between characteristic eye movements and degree of cognitive effort have also enhanced our appreciation of the learning process.
Using linear discriminate analysis (LDA) and nonlinear neural network analyses (NNA) to classify surgeons into expert and nonexpert cohorts, we examine the relationship between complex eye and pupillary movements, collectively referred to as eye metrics, and surgical skill level.
Twenty-one surgeons participated in the simulated and live surgical environments. In the simulated surgical setting, LDA and NNA were able to correctly classify surgeons as expert or nonexpert with 91.9% and 92.9% accuracy, respectively. In the live operating room setting, LDA and NNA were able to correctly classify surgeons as expert or nonexpert with 81.0% and 90.7% accuracy, respectively.
We demonstrate, in simulated and live-operating environments, that eye metrics can reliably distinguish nonexpert from expert surgeons. As current medical educators rely on subjective measures of surgical skill, eye metrics may serve as the basis for objective assessment in surgical education and credentialing in the future. Further development of this potential educational tool is warranted to assess its ability to both reliably classify larger groups of surgeons and follow progression of surgical skill during postgraduate training.
目前，手术技能评估几乎完全依赖于主观测量，而这些测量容易受到多种偏差的影响。我们研究了眼动指标作为评估手术技能的客观工具的应用。
眼动追踪已经帮助阐明了眼动、视觉注意和洞察力之间的关系，所有这些都在执行复杂任务时使用（Kowler 和 Martins，《科学》，1982 年；215:997-999；Tanenhaus 等人，《科学》，1995 年；268:1632-1634；Thomas 和 Lleras，《心理科学述评》，2007 年；14:663-668；Thomas 和 Lleras，《认知》，2009 年；111:168-174；Schriver 等人，《人类因素》，2008 年；50:864-878；Kahneman，《注意与努力》，1973 年）。发现特征性眼动与认知努力程度之间的关联也增强了我们对学习过程的认识。
使用线性判别分析（LDA）和非线性神经网络分析（NNA）将外科医生分为专家和非专家队列，我们研究了复杂的眼动和瞳孔运动（统称为眼动指标）与手术技能水平之间的关系。
21 名外科医生参与了模拟和现场手术环境。在模拟手术环境中，LDA 和 NNA 分别能够以 91.9%和 92.9%的准确率正确地将外科医生分类为专家或非专家。在现场手术室环境中，LDA 和 NNA 分别能够以 81.0%和 90.7%的准确率正确地将外科医生分类为专家或非专家。
我们在模拟和现场手术环境中证明，眼动指标可以可靠地区分非专家和专家外科医生。由于目前的医学教育者依赖于手术技能的主观测量，眼动指标可能成为未来手术教育和认证的客观评估基础。为了评估该潜在教育工具可靠地对更大群体的外科医生进行分类并跟踪其在研究生培训期间手术技能的进展的能力，进一步开发该工具是必要的。
