AI Tools Pose Risks in Clinical Trial Matching, Requiring Human Oversight
The increasing accessibility of AI tools like Microsoft Copilot and ChatGPT is transforming clinical trial matching by making the process more efficient. AI can identify potential study participants from various data sources, including medical records and clinical notes, significantly speeding up patient pre-screening. However, experts like Joshua J. Cook and William 'Billy' Gonzalez Barnes from the University of West Florida emphasize that AI use in this domain carries substantial risks, including potential for inaccurate outputs, unverified information, and threats to patient privacy. They stress the necessity of practical safeguards such as PHI-removal checks, careful prompt preparation, and mandatory human review of AI outputs to mitigate these risks. Standardized quality checks and proper documentation are also highlighted as essential for responsible AI integration in clinical research.