Natural Language Processing and Automated Speech Recognition to Identify Older Adults with Cognitive Impairment Supplement

研究和/或开发用于健康的人工智能技术的现有伦理框架

Existing ethical frameworks for the study and or development of artificial intelligence (AI) technology for health care applications are largely conceptual, lacking critical insights from patients and clinicians, including patients from racial and ethic minority groups. Building on the Parent Study (R01 AG066471), which is developing automated techniques for cognitive impairment screening in primary care, we plan to use qualitative methods to collect perspectives from diverse patient groups, and from physicians, and integrate these data with concepts and data from the literature on ethical challenges of AI research in healthcare. While the Parent Study is specific to cognitive impairment screening, we anticipate that the activities of this project will generate a framework with broader applications. The proposed project harnesses a collaboration of a multi-disciplinary team of clinicians, computer scientists, experts in minority health, and bioethics. The specific aims are (1) to identify and characterize the perceptions and concerns of patients, including those from underrepresented minority groups, and clinicians about AI methods for automated screening for cognitive impairment in outpatient clinical settings; and (2) to integrate this knowledge with existing ethical frameworks of AI research and healthcare screening to develop a more comprehensive ethical framework for the ethical conduct of AI research in healthcare settings. We will first integrate several established ethical frameworks on AI research and healthcare screening and use the resultant preliminary framework to inform qualitative data collection. Next, we will conduct qualitative interviews with patients from diverse backgrounds to understand their perspectives on automated screening and AI research in clinical care (e.g., informed consent, disclosure of results), and focus groups with clinicians for their views on ethical challenges that could hinder adoption. In parallel, we will conduct a PRISMA-Scoping review to identify additional relevant frameworks and research, and interactively refine our qualitative data collection. Finally, we will integrate the qualitative data with concepts from existing literature and frameworks to establish a more comprehensive and inclusive framework than those currently in publication. In this fashion, the work will inform the ethical conduct of research on AI-driven automated cognitive impairment screening and AI research for other conditions. This supplement is responsive to NOT-OD-22-065 by supporting a new collaboration on AI research ethics and developing generalizable methods of exploring and addressing ethical impacts throughout the AI research cycle, specifically through advancing AI research ethical frameworks.