In a landmark policy effort, the American College of Gastroenterology (ACG) published a modified Delphi consensus outlining how artificial intelligence should be responsibly integrated into GI, hepatology, and endoscopy practice.
The paper — Artificial Intelligence for Gastroenterology Practice: A Modified Delphi Consensus — reflects input from 32 experts and 12 industry partners across endoscopy, clinical operations, IBD, liver disease, education, and ethics.
What Reached Consensus
Endoscopy
- Strong evidence supports computer-aided detection (CADe) for improving adenoma detection rate and reducing miss rates.
- Real-world impact remains variable.
- Long-term outcomes (e.g., interval colorectal cancer reduction) are still unproven.
Clinical Operations
- High-potential applications include:
- Ambient AI scribes
- NLP-assisted coding
- Workflow optimization
- Prior authorization automation
- Ambient AI scribes
IBD & Liver Disease
- AI may enhance diagnostic precision.
- Predictive models could improve risk stratification and therapy guidance.
Training & Workforce
- Structured AI education is recommended.
- Independent clinical competence must be preserved to prevent “deskilling.”
Governance & Equity
- Emphasis on:
- Bias mitigation through heterogeneous datasets
- Chain-of-custody data protections
- Specialty-society oversight
- Payment models tied to quality and cost reduction
- Bias mitigation through heterogeneous datasets
The Strategic Signal
This is more than a technology endorsement — it is a governance framework.
ACG positions AI as augmentative, not autonomous, reinforcing that clinical expertise remains central while AI should enhance safety, transparency, interoperability, and equity.