Healthcare AI: Balancing safety and innovation
In this month’s issue of GI & Hepatology News, we summarize a recent systematic review and meta-analysis highlighting worsening health disparities for Hispanic adults with MASLD.
-
01/16/2025
Artificial intelligence (AI) applications are expanding rapidly in healthcare. AI powered tools are increasingly used in everyday medical practice, assisting clinicians with tasks such as diagnosis, treatment planning, data analysis, and patient monitoring, effectively integrating AI into routine clinical decision making. Despite its potential to fundamentally transform the practice of medicine and healthcare delivery, AI in healthcare remains largely unregulated, with a lack of common standards to guide responsible design, development, and adoption of AI-based tools to guide clinical care.
But this is changing. In mid-January, the US Department of Health & Human Services released its Strategic Plan for the Use of AI in Health, Human Services, and Public Health (available at www.healthit.gov), presenting an approach to catalyze innovation, promote trustworthy AI development, democratize technologies and resources, and cultivate AI-empowered workforces and organizational cultures. While there is no immediate regulatory impact, the plan does provide important insights into how the federal government thinks about AI, which will be a part of driving regulations in the future. As crucial stakeholders in the health AI universe and advocates for its responsible use in clinical practice, it is critical that we as clinicians keep abreast of developments in this rapidly evolving space.
In this month’s issue of GI & Hepatology News, we summarize a recent systematic review and meta-analysis highlighting worsening health disparities for Hispanic adults with MASLD. We also report the results of an industry-sponsored study comparing the real-world clinical effectiveness of GI Genius (an AI-driven tool) with that of standard colonoscopy.
In February’s Member Spotlight, we introduce you to international AGA member Dr. Tossapol Kerdsirichairat (clinical associate professor of gastroenterology at Bumrungrad International Hospital in Bangkok, Thailand), who shares his insights regarding the challenges and rewards of practicing gastroenterology at one of the largest private hospitals in Southeast Asia. ‘Tos’ is one of roughly 25% of AGA members who live and work outside the United States.
Finally, this month’s In Focus column from The New Gastroenterologist focuses on management of chronic constipation, a highly prevalent condition that significantly impacts the quality of life of many of our patients. We hope you enjoy this and all the exciting content in our February issue.
Megan A. Adams, MD, JD, MSc
Editor in Chief
Summary content
7 Key Takeaways
-
1
Developed a paper-based colorimetric sensor array for chemical threat detection.
-
2
Can detect 12 chemical agents, including industrial toxins.
-
3
Production cost is under 20 cents per chip.
-
4
Utilizes dye-loaded silica particles on self-adhesive paper.
-
5
Provides rapid, simultaneous identification through image analysis.
-
6
Inspired by the mammalian olfactory system for pattern recognition.
-
7
Future developments include a machine learning-enabled reader device.
The guidelines emphasize four-hour gastric emptying studies over two-hour testing. How do you see this affecting diagnostic workflows in practice?
Dr. Staller: Moving to a four-hour solid-meal scintigraphy will actually simplify decision-making. The two-hour reads miss a meaningful proportion of delayed emptying; standardizing on four hours reduces false negatives and the “maybe gastroparesis” purgatory that leads to repeat testing. Practically, it means closer coordination with nuclear medicine (longer slots, consistent standardized meal), updating order sets to default to a four-hour protocol, and educating front-line teams so patients arrive appropriately prepped. The payoff is fewer equivocal studies and more confident treatment plans.
Metoclopramide and erythromycin are the only agents conditionally recommended for initial therapy. How does this align with what is being currently prescribed?
Dr. Staller: This largely mirrors real-world practice. Metoclopramide remains the only FDA-approved prokinetic for gastroparesis, and short “pulsed” erythromycin courses are familiar to many of us—recognizing tachyphylaxis limits durability. Our recommendation is “conditional” because the underlying evidence is modest and patient responses are heterogeneous, but it formalizes what many clinicians already do: start with metoclopramide (lowest effective dose, limited duration, counsel on neurologic adverse effects) and reserve erythromycin for targeted use (exacerbations, bridging).
Several agents, including domperidone and prucalopride, received recommendations against first-line use. How will that influence discussions with patients who ask about these therapies?
Dr. Staller: Two points I share with patients: evidence and access/safety. For domperidone, the data quality is mixed, and US access is through an FDA IND mechanism; you’re committing patients to EKG monitoring and a non-trivial administrative lift. For prucalopride, the gastroparesis-specific evidence isn’t strong enough yet to justify first-line use. So, our stance is not “never,” it’s just “not first.” If someone fails or cannot tolerate initial therapy, we can revisit these options through shared decision-making, setting expectations about benefit, monitoring, and off-label use. The guideline language helps clinicians have a transparent, evidence-based conversation at the first visit.
The guidelines suggest reserving procedures like G-POEM and gastric electrical stimulation for refractory cases. In your practice, how do you decide when a patient is “refractory” to medical therapy?
Dr. Staller: I define “refractory” with three anchors.
1. Adequate trials of foundational care: dietary optimization and glycemic control; an antiemetic; and at least one prokinetic at appropriate dose/duration (with intolerance documented if stopped early).
2. Persistent, function-limiting symptoms: ongoing nausea/vomiting, weight loss, dehydration, ER visits/hospitalizations, or malnutrition despite the above—ideally tracked with a validated instrument (e.g., GCSI) plus nutritional metrics.
3. Objective correlation: delayed emptying on a standardized 4-hour solid-meal study that aligns with the clinical picture (and medications that slow emptying addressed).
At that point, referral to a center with procedural expertise for G-POEM or consideration of gastric electrical stimulation becomes appropriate, with multidisciplinary evaluation (GI, nutrition, psychology, and, when needed, surgery).
What role do you see dietary modification and glycemic control playing alongside pharmacologic therapy in light of these recommendations?
Dr. Staller: They’re the bedrock. A small-particle, lower-fat, calorie-dense diet—often leaning on nutrient-rich liquids—can meaningfully reduce symptom burden. Partnering with dietitians early pays dividends. For diabetes, tighter glycemic control can improve gastric emptying and symptoms; I explicitly review medications that can slow emptying (e.g., opioids; consider timing/necessity of GLP-1 receptor agonists) and encourage continuous glucose monitor-informed adjustments. Pharmacotherapy sits on top of those pillars; without them, medications will likely underperform.
The guideline notes “considerable unmet need” in gastroparesis treatment. Where do you think future therapies or research are most urgently needed?
Dr. Staller: I see three major areas.
1. Truly durable prokinetics: agents that improve emptying and symptoms over months, with better safety than legacy options (e.g., next-gen motilin/ghrelin agonists, better-studied 5-HT4 strategies).
2. Endotyping and biomarkers: we need to stop treating all gastroparesis as one disease. Clinical, physiologic, and microbiome/omic signatures that predict who benefits from which therapy (drug vs G-POEM vs GES) would transform care.
3. Patient-centered trials: larger, longer RCTs that prioritize validated symptom and quality-of-life outcomes, include nutritional endpoints, and reflect real-world medication confounders.
Our guideline intentionally highlights these gaps to hopefully catalyze better trials and smarter referral pathways.
Dr. Staller is with the Division of Gastroenterology, Massachusetts General Hospital and Harvard Medical School, Boston.