The Era of Generative Artificial Intelligence in Clinical Care: A Case Study in Acute Gastrointestinal Bleeding

May 1, 2025
Medical Report

The Era of Generative Artificial Intelligence in Clinical Care: A Case Study in Acute Gastrointestinal Bleeding

Dennis L. Shung, MD, MHS, PhD

This article by Dr. Dennis Shung, an Assistant Professor of Medicine and Director of Digital Health at Yale School of Medicine, explores how generative AI can transform care for patients with acute gastrointestinal bleeding. A leading expert at the intersection of AI and clinical medicine, Dr. Shung illustrates how tools like large language models and predictive algorithms can improve early detection, risk assessment, and decision-making. His case study reflects a broader trend-AI is rapidly reshaping medical workflows and empowering clinicians to deliver smarter, more efficient, and more personalized care.

Generative artificial intelligence will transform the practice of medicine by disrupting existing workflows and changing the dynamic of patient-physician interactions. Multimodal data generated during routine clinical care can increasingly be used to enhance patient care. Gastroenterology and hepatology is a specialty that routinely generates and synthesizes multimodal data for patient disease management, and is on the forefront of artificial intelligence (AI) integration in routine clinical care through computer-aided polyp detection (CADe) systems during screening and surveillance colonoscopy. However, AI will permeate and change all aspects of future clinical workflows, and not just during endoscopic procedures. As a case study, I will explore the potential impact of AI on the most common cause for GI-related hospitalization, acute gastrointestinal bleeding (GIB).

For context, acute GIB accounts for over 27.7 billion dollars over 475,000 hospitalizations annually in the United States.[1] Risk stratification to identify very low risk patients is recommended by national and international guidelines.[2-6] Machine learning models to predict risk of hospital-based intervention outperform existing clinical risk scores in identifying very low risk patients who can be discharged for outpatient management.[7] Identifying and discharging low risk patients with machine learning algorithms can capture potential cost savings of 3.4 billion for upper gastrointestinal bleeding alone.[8]

AI via local large language models can identify patients via named entity recognition and electronic health record phenotyping with very high positive predictive value who have signs or symptoms of acute gastrointestinal bleeding. In a study that evaluated the performance of identifying signs or symptoms of acute gastrointestinal bleeding from nursing notes, a customized prompt architecture for a local language model LLaMA-2-70B had PPV of 97%.[9] Once patients are automatically identified, a trained electronic health record-based machine learning model that outperforms the Glasgow Blatchford Score and Oakland Score can be automatically deployed with a recommendation at the high sensitivity threshold (pre-defined as >99%).[10] This predictive model output and interpretation can be then presented along with clinical guideline-driven decision support via a chatbot interface, where physician users are able to query both the justification for the risk and ask any questions pertaining to the management of patients with acute gastrointestinal bleeding.[11, 12] If the patient is admitted for observation or inpatient hospitalization, generative AI will track the dynamic risk over time by predicting the hemodynamic trajectories.[13] If the patient undergoes diagnostic upper endoscopy and an ulcer is noted, convoluted neural network-based algorithms can evaluate for high risk features.[14] Throughout, certain ranges where there exists greater uncertainty from the algorithmic perspective will prompt the human user to seek additional expertise to prevent overconfidence in algorithmic outputs.[15]

As the proposed workflow suggests, AI could enhance the clinical care of patients with acute gastrointestinal bleeding to promptly identify, risk stratify, follow, and support clinical management pre- endoscopically and during endoscopy. In the future when these platforms are deployed at scale, the reality may be that while good AI can’t replace a bad clinician, good AI can empower clinicians to be better.

References

Peery AF, Murphy CC, Anderson C, et al. Burden and Cost of Gastrointestinal, Liver, and Pancreatic Diseases in the United States: Update 2024. Gastroenterology 2025.
Laine L, Barkun AN, Saltzman JR, et al. ACG Clinical Guideline: Upper Gastrointestinal and Ulcer Bleeding. Am J Gastroenterol 2021;116:899-917.
Barkun AN, Almadi M, Kuipers EJ, et al. Management of Nonvariceal Upper Gastrointestinal Bleeding: Guideline Recommendations From the International Consensus Group. Ann Intern Med 2019;171:805-822.
Gralnek IM, Stanley AJ, Morris AJ, et al. Endoscopic diagnosis and management of nonvariceal upper gastrointestinal hemorrhage (NVUGIH): European Society of Gastrointestinal Endoscopy (ESGE) Guideline – Update 2021. Endoscopy 2021;53:300-332.
Sung JJ, Chiu PW, Chan FKL, et al. Asia-Pacific working group consensus on non-variceal upper gastrointestinal bleeding: an update 2018. Gut 2018;67:1757-1768.
Oakland K, Chadwick G, East JE, et al. Diagnosis and management of acute lower gastrointestinal bleeding: guidelines from the British Society of Gastroenterology. Gut 2019;68:776-789.
Shung D, Simonov M, Gentry M, et al. Machine Learning to Predict Outcomes in Patients with Acute Gastrointestinal Bleeding: A Systematic Review. Dig Dis Sci 2019;64:2078-2087.
Shung DL, Lin JK, Laine L. Achieving Value by Risk Stratification With Machine Learning Model or Clinical Risk Score in Acute Upper Gastrointestinal Bleeding: A Cost Minimization Analysis. Am J Gastroenterol 2024;119:371-373.
Zheng NS, Keloth VK, You K, et al. Detection of Gastrointestinal Bleeding With Large Language Models to Aid Quality Improvement and Appropriate Reimbursement. Gastroenterology 2025;168:111-120. e4.
Shung DL, Chan CE, You K, et al. Validation of an Electronic Health Record-Based Machine Learning Model Compared With Clinical Risk Scores for Gastrointestinal Bleeding. Gastroenterology 2024.
Chan C, You K, Chung S, et al. Assessing the Usability of GutGPT: A Simulation Study of an AI Clinical Decision Support System for Gastrointestinal Bleeding Risk, 2023:arXiv:2312.10072.
Rajashekar NC, Shin YE, Pu Y, et al. Human- Algorithmic Interaction Using a Large Language Model-Augmented Artificial Intelligence Clinical Decision Support System. Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems. Honolulu, HI, USA: Association for Computing Machinery, 2024:Article 442.
Zhang X, Pu Y, Kawamura Y, et al. Trajectory Flow Matching with Applications to Clinical Time Series Modeling, 2024:arXiv:2410.21154.
He XJ, Wang XL, Su TK, et al. Artificial intelligence- assisted system for the assessment of Forrest classification of peptic ulcer bleeding: a multicenter diagnostic study. Endoscopy 2024;56:334-342.
Alur R, Laine L, Li DK, et al. Auditing for human expertise. Proceedings of the 37th International Conference on Neural Information Processing Systems. New Orleans, LA, USA: Curran Associates Inc., 2024:Article 3477.

Dennis L. Shung, MD, MHS, PhD

Assistant Professor of Medicine (Digestive Diseases)
Yale School of Medicine

Dennis L. Shung, MD, MHS, PhD is an Assistant Professor of Medicine and Director of Digital Health in the Section of Digestive Diseases at Yale School of Medicine, with a secondary appointment in Biomedical Informatics. He also leads applied AI efforts at Yale’s Healthcare Simulation Center and Clinical and Translational Research Accelerator.

Dr. Shung is the founder of Yale’s Human+Artificial Intelligence in Medicine (H+AIM) Lab, which focuses on building trust and value in human-AI collaboration. He contributes to national AI policy through the Coalition for Health AI and holds leadership roles in key GI and AI professional organizations. A physician data scientist, his work spans clinical care, machine learning, and implementation science, with publications in Gastroenterology, Nature, and JAMA Network.

In August 2025, Dr. Shung will join the Mayo Clinic as Associate Professor of Medicine, Physician Lead for Digital Innovation in the Department of Medicine, and Director of Clinical Generative AI and Informatics, where he will continue advancing the integration of artificial intelligence into patient care and health systems.

Previous Posts Patient Care Should Take Precedence Over Any Artificial Dilemmas Imposed by Regulatory Bodies

Next Posts COVID/Pandemic and Biopsychosocial Impacts on Children

Kyoung Ryul Lee, MD, PhD, CEO of the SCL Group, Seoul Clinical Laboratory, a Specialist in Laboratory Medicine

ISSUE 28

Youngmee Jee, MD, PhD | Inspirational Asian Healthcare Leader

Rethinking Hypothyroidism

Upcoming Conference

The Internet of Things (Revision 2021)

Superconvergence

The Strategic Importance of Location in the U.S. for Foreign Bio and Pharma Companies

South Korea’s Healthcare Crisis: Beyond the Numbers

Overcoming Disparities in Gastric Cancer Care

The Era of Generative Artificial Intelligence in Clinical Care: A Case Study in Acute Gastrointestinal Bleeding

References

Dennis L. Shung, MD, MHS, PhD

Leave Your Comment

Quick Link

Get us in your inbox