 of colorectal polyps, are already being used in clinical practice. These technologies help identify polyps during endoscopy, supporting clinicians in making more accurate diagnoses.){kind=link}
AI algorithms, particularly for computer-aided detection (CADe) of colorectal polyps, are already being used in clinical practice. These technologies help identify polyps during endoscopy, supporting clinicians in making more accurate diagnoses. Beyond CADe, AI applications in GI endoscopy are rapidly advancing, with significant interest in computer-aided diagnosis (CADx) systems that could potentially replace traditional histopathology for diagnosing small colon polyps. This could support “resect and discard” or “diagnose and leave” strategies, streamlining procedures and reducing costs.
Another promising area for AI integration is capsule endoscopy, where fully automated reporting may soon become a reality. These advancements could lead to unprecedented levels of efficiency in GI endoscopy, yet they also raise critical questions about errors and liability. The degree of legal responsibility for errors made by AI algorithms depends on the level of automation and how these tools are integrated into clinical practice.
To understand the legal ramifications, it is essential to distinguish between different levels of AI automation, ranging from limited decision support to full automation. For instance, the U.S. Food and Drug Administration has approved several CADe tools that operate at a basic assistive level, where clinicians can choose to respond to or ignore AI alerts. Enthusiasm for CADe technology is high, yet its adoption in clinical practice has been slower than anticipated, potentially due to implementation costs.
The next wave of AI in GI endoscopy is likely to include “level 2” CADx tools, which provide predictive diagnoses for polyps, augmenting clinicians’ optical assessments. These tools could reduce the need for histopathology, potentially lowering costs and avoiding unnecessary procedures. However, only 57% of gastroenterologists are currently willing to consider CADx-supported strategies for small polyps, citing concerns about liability for diagnostic errors.
Future AI developments may introduce even higher levels of automation, particularly in capsule endoscopy. Current software already uses indicators to highlight areas of interest, but upcoming AI-driven algorithms could detect and diagnose a broader range of pathologies with minimal clinician intervention. These advancements could range from preselecting high-value images for review to fully automating the entire interpretation process.
As AI continues to play a more prominent role in GI endoscopy, addressing the liability implications is crucial. Physicians and healthcare organizations must carefully consider how to integrate these tools to balance the benefits of AI with the potential risks and legal responsibilities.