ASGE Postgraduate Course at ACG: Innovative Practices for the Endoscopist of Today, Tomorrow and Beyond (On-demand)

Pdf Summary

Dr. Folasade P. May's presentation addresses the significant role artificial intelligence (AI) can play in mitigating health disparities in gastroenterology. She begins by explaining health equity and the importance of providing everyone with fair opportunities to achieve optimal health, emphasizing that health disparities often stem from economic, social, or environmental disadvantages.<br /><br />The presentation highlights prevalent biases in AI applications within gastroenterology:<br />1. <strong>Esophageal Cancer:</strong> Research predominantly focuses on early detection technologies for Barrett's esophagus and esophageal adenocarcinoma (EAC), mostly benefiting white populations despite esophageal squamous cell carcinoma (ESCC) being more common globally and disproportionately affecting Black and Asian individuals, indicating selection bias.<br />2. <strong>Inflammatory Bowel Disease (IBD):</strong> Many AI/ML algorithms are trained predominantly on white populations, despite an increasing incidence of IBD among non-white groups. This data collection bias can lead to poorer management and outcomes for Black and Latino patients.<br />3. <strong>Colorectal Cancer:</strong> While AI assists in detecting high-risk polyps via colonoscopy, underserved and vulnerable populations often lack access to these technologies, further perpetuating disparities owing to algorithm development and post-deployment biases.<br /><br />To address these discrepancies, Dr. May proposes several solutions:<br />- <strong>Incorporate Health Equity Experts:</strong> Engage experts in the development and deployment stages of AI.<br />- <strong>Diversify Study Populations and Settings:</strong> Ensure training datasets represent marginalized populations and expand research locations to diverse community settings.<br />- <strong>Implement Regulatory Measures:</strong> Establish fair, clear, and specific measures to evaluate outcomes and mandate reporting of study population demographics.<br />- <strong>Conduct Pre- and Post-Deployment Auditing:</strong> Assess algorithmic performance across subpopulations and establish processes to promptly correct biases.<br /><br />In summary, while AI presents valuable opportunities for advancing medical science and healthcare, proactively addressing biases is crucial to prevent perpetuating or exacerbating health inequities.

Meta Tag

Concept Health Equity

Concept Health Disparity

Concept Artificial Intelligence

Concept Bias

Concept Selection Bias

Keywords

Artificial Intelligence

Health Disparities

Gastroenterology

Health Equity

Esophageal Cancer

Inflammatory Bowel Disease

Colorectal Cancer

Bias Mitigation

Diverse Populations

Regulatory Measures

Health Equity

Health Disparity

Artificial Intelligence

Bias

Selection Bias