Benway Nina R, Preston Jonathan L
Department of Electrical and Computer Engineering, University of Maryland, College Park.
Department of Communication Sciences & Disorders, Syracuse University, NY.
Lang Speech Hear Serv Sch. 2025 Jul 16;56(3):442-468. doi: 10.1044/2025_LSHSS-24-00085. Epub 2025 May 15.
Artificial intelligence (AI) is more capable and accessible than ever before. But what does this mean for clinical practice? How can speech-language clinicians evaluate the efficacy, validity, and reliability of AI and machine learning tools for automating assessment and treatment? How can speech-language clinicians ethically use these clinical AI technologies? We contend that clinical AI will best serve clinicians and clients when aligned with an evidence-based framework. Therefore, this tutorial presents guidelines for the critical appraisal of clinical AI through the lens of validity, reliability, ethical use, and equitable use, facilitated by the Critical Appraisal Rubric for Ethical and Equitable Clinical Artificial Intelligence. Similarly, in order for developers of clinical AI to meet the needs of the profession, these principles should guide the development and assessment of new clinical technologies.
The questions of efficacy, validity, reliability, ethical use, and equitable use of clinical AI can be answered through the examination of a specific clinical AI for a given user, as emphasized by culturally responsive professional practice. A framework is provided to assist clinicians in the critical appraisal of clinical AI tools.
人工智能(AI)比以往任何时候都更具能力且更易获取。但这对临床实践意味着什么?言语语言治疗师如何评估用于自动化评估和治疗的人工智能及机器学习工具的疗效、有效性和可靠性?言语语言治疗师如何在道德层面使用这些临床人工智能技术?我们认为,当临床人工智能与基于证据的框架保持一致时,它将最有利于临床医生和患者。因此,本教程通过《道德与公平临床人工智能批判性评估量表》,从有效性、可靠性、道德使用和公平使用的角度,介绍临床人工智能批判性评估的指导原则。同样,为了使临床人工智能开发者满足该专业的需求,这些原则应指导新临床技术的开发和评估。
正如具有文化适应性的专业实践所强调的,通过针对特定用户检查特定的临床人工智能,可以回答临床人工智能的疗效、有效性、可靠性、道德使用和公平使用等问题。本文提供了一个框架,以协助临床医生对临床人工智能工具进行批判性评估。