Introduction

Upper gastrointestinal bleeding (UGIB) remains a common and potentially life-threatening condition in emergency medicine. Early identification of patients requiring urgent esophagogastroduodenoscopy (EGD) is crucial to improve outcomes and optimize healthcare resources management. Traditional risk stratification tools, such as the Glasgow-Blatchford Score (GBS), have been widely used to guide clinical decision-making. However, the growing availability of artificial intelligence (AI) offers new perspectives in diagnostic and therapeutic strategies. Recent advancements in large language models (LLMs), such as ChatGPT® and Google GEMINI®, suggest potential applications in clinical triage.

Aims & Methods

This study aims to evaluate the performance of these AI tools in determining the appropriate timing for endoscopy in UGIB patients, compared to standard clinical scoring systems.
The medical records of 100 patients who presented to the emergency department with upper gastrointestinal bleeding (UGIB) between April and September 2024 were reviewed retrospectively. Each case was evaluated by two AI models: Google GEMINI® (September 2024 version) and ChatGPT® (version 3.5). These models, acting as virtual gastroenterology experts, were asked to recommend whether urgent or delayed esophagogastroduodenoscopy (EGD) was necessary. For each patient, clinical data such as age, medical history, clinical examination findings, and laboratory results at admission were provided. The AI recommendations were recorded in an Excel file and compared with the Glasgow-Blatchford Score (GBS), which was calculated by a gastroenterologist at the time of admission.
An urgent endoscopy was deemed necessary for a GBS ≥ 3. Statistical analysis was performed using Jamovi® software (version 2.3), and the agreement between the AI models and human evaluations was assessed using Cohen’s kappa coefficient (κ).

Results

The epidemiological characteristics of the patient cohort revealed a median age of 63 years (range: 37.5 to 78), with a male-to-female ratio of 1.7. Among the patients, 73% had at least one comorbidity, including hypertension (33%), diabetes (22%), ischemic heart disease (17%), and chronic liver disease (12%). At the time of admission, melena was the predominant symptom, reported in 78% of patients. The average Glasgow-Blatchford Score (GBS) was 7.07 ± 3.47.
In terms of AI recommendations, ChatGPT suggested urgent endoscopy in 48% of cases, while Google GEMINI recommended it in 49%. A moderate level of agreement was observed between the two AI models for urgent endoscopy recommendations (kappa = 0.540, p < 0.001). However, the agreement between the AI models and the human evaluator was weak. The level of agreement between the gastroenterologist’s assessment and ChatGPT’s recommendation was weak (kappa = 0.222, p < 0.001), with similar results for Google GEMINI (kappa = 0.231, p < 0.001).

Conclusion

Artificial intelligence is a promising tool that is already beginning to transform various fields of medicine. However, our study demonstrates that the publicly available AI models currently in use are not yet capable of effectively triaging patients in emergency settings, particularly in cases of upper gastrointestinal bleeding. A model specifically trained on clinical gastroenterology cases, using validated scientific guidelines, could produce more accurate results and potentially revolutionize patient triage in emergency departments.

Introduction

Endoscopy is essential for the early detection and management of upper gastrointestinal (UGI) tumors, yet variability in diagnostic expertise and workload-related fatigue often compromise the quality of report documentation. These challenges underscore the urgent need for innovative technological approaches to assist endoscopists in optimizing the report generation process.
Multimodal large language models (MLLMs) enable the synergistic integration of visual and semantic representations, demonstrating significant clinical value in medical text generation. This study introduces MERGS, an innovative endoscopy report generation system powered by MLLM. MERGS annotates medical terminology and generates natural language descriptions based on endoscopic images, ultimately producing structured reports that conform to clinical standards.

Aims & Methods

MERGS is built on the pre-trained MLLM Qwen-VL-2.5-7B. It adopts a multimodal architecture integrating visual encoding and cross-modal alignment to enable medical term annotation and natural language generation from endoscopic images. A pathological classification module further enhances lesion characterization. The language model synthesizes image-derived descriptions to generate a comprehensive report with a diagnostic conclusion.
UGI endoscopy reports were collected from the Sun Yat-sen University Cancer Center between 2019 and 2021. A standardized terminology dictionary was developed based on clinical guidelines, supporting a three-tier annotation framework: endoscopic image → standardized medical terminology → natural language text. Model-assisted pre-annotation combined with expert validation substantially reduced manual effort. The resulting dataset was split into training and test sets at an 8:2 ratio.
Model performance was assessed via multi-label classification metrics (accuracy, specificity, F1-score), as well as semantic similarity measures (Jaccard index) to evaluate its ability to recognize anatomical sites and endoscopic features. Natural language generation quality was assessed using BLEU-2 and ROUGE-L scores, and expert satisfaction ratings were collected for 100 randomly sampled outputs to assess readability and clinical relevance. A human–AI comparison was conducted on 50 test reports, comparing system-generated outputs with those authored by experienced endoscopists.

Results

A multimodal dataset was constructed after quality control, comprising 22,373 reports and 135,503 associated images from 21,619 patients. Internal validation demonstrated that MERGS achieved robust performance in automated image interpretation, with 97.3% accuracy and 98.6% specificity for standard anatomical site recognition. In feature identification, it showed high clinical relevance (F1 scores: smooth mucosa 0.915, polyp 0.857, tumor 0.878, erythematous mucosa 0.766).
Generated reports showed strong alignment with reference reports (BLEU-2 = 0.587, ROUGE-L = 0.661) and received high expert satisfaction ratings (mean score: 4.34 ± 0.63 on a 5-point scale). In a human–AI comparison, MERGS-generated reports were considered non-inferior to those authored by endoscopists (scores: MERGS 4.52 ± 0.81 vs. manual 4.68 ± 0.72), with the upper limit of the one-sided 95% CI (0.37) below the non-inferiority margin (0.5, p = 0.0045).

Conclusion

This study developed an automated system for UGI endoscopy reporting powered by an MLLM. Both internal and clinical validation demonstrated the system's high performance in endoscopic image feature recognition and report generation, underscoring its strong potential for clinical deployment.

References

Luo H, Xu G, Li C, He L, Luo L, Wang Z, Jing B, Deng Y, Jin Y, Li Y, Li B, Tan W, He C, Seeruttun S R, Wu Q, Huang J, Huang D W, Chen B, Lin S B, Chen Q M, Yuan C M, Chen H X, Pu H Y, Zhou F, He Y, Xu R H. Real-time artificial intelligence for detection of upper gastrointestinal cancer by endoscopy: a multicentre, case-control, diagnostic study[J]. Lancet Oncol, 2019,20(12):1645-1654.

Introduction

Pre-cut sphincterotomy are most often used as a salvage procedure after failure of conventional methods of bile duct cannulation although its success is not guaranteed.　
When we performed initial pre-cut sphincterotomy, papilla is incited, with needle knife or sphincterotome, until the opening of bile duct is recognized to gain access to its lumen.
Our technique, in cases of not recognized the opening of bile duct, involves engaging the exposing underlying submucosal tissue layer of the papilla with the catheter and then seeking bile duct with the hydrophilic guide wire, completely dependent on fluoroscopic view.
The purpose of this study was to assess the efficacy and safety of our technique in cases of failed pre-cut.

Aims & Methods

We retrospectively analyzed the clinical records of consecutive patients undergoing ERCP between 1 January 2009 and 31 October 2024 at Mitsui Memorial Hospital, Tokyo, Japan.
Patients who underwent an ERCP plus pre-cut were included.
Patients were divided into two groups, in whom the common bile duct could be cannulated (Group A), and the common bile duct could not be cannulated and our technique was performed(Group B), after initial pre-cut.
The primary objective was a successful biliary cannulation, and the secondary objective was the complication in group B.
The complications and patient characteristics were compared between group A and B using Fisher’s exact test.

Results

Over a period of fourteen years ten months, 262 Pre-cut sphincterotomy (Needle knife sphincterotomy 108, Transpancreatic septotomy 126, Both 28) were performed. After initial pre-cut, we could achieve biliary access in 183 (Group A), and could not in 79. We performed our technique in 77 among these 79(group B).
In Group B, the therapeutic indications for ERCP were choledocholithiasis in 44 (57.1%) of cases, malignant and benign biliary obstruction in 29(37.7%) and 4(5.2%). The average age of the patients was 73 years (range 29-96) and 45 (58.4%) were male. Among 77 patients, a successful biliary cannulation was achieved in 72 (93.5 %). Complications were 15(19.5%). 2 patients presented bleeding (2.6 %) and required endoscopic treatment and transfusion of red blood cells. 9 (11.7 %) patients developed post-ERCP pancreatitis (1 severe, 8 mild pancreatitis). Guidewire related perforations that were resolved with conservative management occurred in 4(5.2%).
Alerted anatomy status was more frequently observed in group B than group A (p < 0.01). The other characteristics of both groups were similar. Total complications were more frequently observed in group B (p ＝ 0.03).

Conclusion

Our technique in cases of failed pre-cut sphincterotomy is found to be highly effective method to get biliary access, but carried a higher adverse event rate.

Introduction

Abdominal pain is a common symptom in clinical settings, but its prevalence and impact on individuals in the global population are poorly understood.

Aims & Methods

We aimed to use the population-based Rome Foundation Global Epidemiology Study (RFGES)¹ dataset to assess the prevalence and burden of abdominal pain in the global population. Data from all 54127 individuals in 26 countries covering 6 world regions (Asia, Middle East, Eastern Europe, Western Europe, North America, and Latin America) surveyed via the Internet (49.1% females; mean age 43.7 years, range 18-97) were included in the analyses. The survey included the Rome IV diagnostic questionnaire, the PROMIS Global-10 quality of life (QoL) questionnaire, the Patient Health Questionnaire PHQ-4 (anxiety and depression), and questions about medical history and healthcare utilization. Survey respondents who reported pain anywhere in the abdomen in the past 3 months were classified as abdominal pain cases, and all others as controls.

Results

Abdominal pain was reported by 51.9% (95% CI: 51.4, 52.3) of people in the global sample. The prevalence was higher among women than men [59.6% (59.0, 60.1) vs. 44.4% (43.8, 45.0)], and declined across age groups: 60.5% (59.8, 61.2) for ages 18 to 34, 54.6% (53.8, 55.4) for ages 35 to 49 years, 45.5% (44.6, 46.4) for 50-64 years, and 37.7% (36.7, 38.7) for 65+ years. Abdominal pain prevalence varied markedly between world regions, from a low of 41.2% (40.2, 42.2) in Asia to a high of 66.5% (65.3, 67.7) in Eastern Europe. Among the 26 individual countries, it ranged from 32.1% (30.2, 33.9) in Japan to 69.3% (67.2, 71.3) in Egypt.
Individuals with abdominal pain were twice as likely as controls to meet criteria for one or more of the 22 disorders of brain-gut interaction (DGBI) assessed [54.2% (53.6, 54.8) vs. 25.0% (24.5, 25.5)] and more than twice as likely to report a history of organic gastrointestinal disorders: [9.8% (9.4, 10.1) vs. 4.2% (3.9, 4.4)]; p<0.0001 for both. However, 38.9% of people with abdominal pain neither met DGBI criteria nor had organic GI disease history.
People with abdominal pain had significantly lower physical and mental quality of life scores than controls, showed more impairment in work productivity and life activities, were more likely to have ever visited a doctor because of a bowel problem, and more likely to visit doctors frequently (at least monthly) for any health reasons. They also more commonly reported being concerned about their bowel functioning. For details see Table 1.

Conclusion

Abdominal pain affects half the adult global population, including many people who have no GI disorders. It is more common in women, younger individuals, and people in non-Asian countries. Having abdominal pain is associated with poorer quality of life, impaired work and life functioning, increased psychological symptoms and greater healthcare utilization. These findings underscore that abdominal pain, as a symptom rather than only as a part of diagnosed disorders, warrants attention by healthcare services and researchers to better address unmet needs of sufferers.

References

1. Sperber et al. Worldwide Prevalence and Burden of Functional Gastrointestinal Disorders, Results of Rome Foundation Global Study. Gastroenterology. 2021 Jan;160(1):99-114.e3. doi: 10.1053/j.gastro.2020.04.014.

Disclosure

BBDF, CA and RL are currently employees of Opella, a Sanofi company and may hold shares or stock options in the company. OSP receives licensing fees from MetaMe Health. MS has received unrestricted research grants from BioGaia; served as consultant/advisory board member for Biocodex, Tillotts, BioGaia, Renapharma, and AlfaSigma; and served on speakers' bureau for Tillotts, Takeda, Biocodex, Sanofi, Abbvie, Janssen Immunology, Pfizer, BioGaia, Renapharma, Mayoly and Bromatech. JT has given scientific advice to Aclipse, Adare, AlfaSigma, Bio-Codex, Clasado, Danone, Falk, FitForMe, Ironwood, Kyowa Kirin, Menarini, Promed, Ricordati, Takeda, Tsumura and Truvion Pharmaceuticals; received research support from Biohit, Kiowa Kirin, ProMed, Sofar, Takeda and Tsumura; and served on the speaker bureau for Abbott, Bio-Codex, Mayoly, Menarini, ProMed, Reckitt-Benckiser, Schwabe, Takeda, Thai Meiji and Truvion Pharmaceuticals. JH, DAD, ADS and SIB have no conflicts of interest to declare.

Introduction

Endoscopic retrograde cholangiopancreatography (ERCP) is a widely utilized procedure for managing biliary and pancreatic disorders (1). ERCP is associated with complications, and potential factors influencing the ERCP outcomes is a critical area of research (2). Among these factors, obesity has emerged as a variable of interest due to its increasing prevalence globally and its known associations with a range of health conditions (3). Previous studies have yielded conflicting results on the influence of obesity on ERCP outcomes, with some reporting higher complication rates, while others found no significant associations (4, 5). The interplay between weight and procedural outcomes across a spectrum of BMI categories remains underexplored. This study aimed to assess the impact of body mass index (BMI) on clinical aspects and procedural outcomes of ERCP.

Aims & Methods

A total of 2143 patients who underwent ERCP at Hitit University Hospital between January 2020 and February 2024 were initially included. Of these, 1315 patients were excluded due to incomplete clinical information, previous ERCP history, altered anatomy and pregnancy, resulting in a final cohort of 828 patients for analysis.
The included patients were stratified into BMI categories: ≤30 kg/m² and >30 kg/m². Additionally, subgroup analyses were performed across finer BMI classifications: ≤20, 20 to ≤25, 25 to ≤30, 30 to ≤35, and >35 kg/m². Clinical and procedural parameters evaluated included admission diagnoses of acute pancreatitis and acute cholangitis (according to Tokyo cholangitis grade), American Society of Anesthesiologists (ASA) scores, gender, history of cholecystectomy, cannulation success, cannulation time, post-ERCP complications including post-ERCP pancreatitis and other complications according to modified Cotton criteria and 30-day readmission rates. The relationships between BMI and the clinical aspects as well as procedural outcomes of ERCP were investigated.

Results

The study cohort consisted of 828 patients, with 75.8% (n=628) having a BMI ≤30 kg/m² and 24.2% (n=200) having a BMI >30 kg/m². Within the finer BMI categories, 3.7% (n=31) of patients had a BMI ≤20 kg/m², 24.2% (n=200) were in the 20 to ≤25 kg/m² range, 47.9% (n=397) were in the 25 to ≤30 kg/m² range, 19.7% (n=163) were in the 30 to ≤35 kg/m² range, and 4.5% (n=37) had a BMI >35 kg/m². The majority of patients (47.9%) were within the BMI range of 25–30 kg/m², reflecting a significant proportion of overweight individuals in the cohort. Patients with a BMI >35 kg/m² constituted the smallest group, accounting for only 4.5% of the total population.
The study found no statistically significant differences between the groups concerning admission rates for acute pancreatitis and acute cholangitis, ASA scores, cannulation success, post-ERCP complications, and readmission within 30 days after the procedure (p>0.05). Notably, women were found to be significantly more obese (p<0.001), and a higher weight was associated with an increased history of cholecystectomy (p=0.013).
Additionally, there were no significant differences in average age, Charlson comorbidity index, cannulation time, length of hospital stay, or white blood cell count at admission between the BMI groups (p>0.05). However, CRP and total bilirubin levels were significantly higher in the non-obese group compared to the obese group (p=0.01 and p=0.011, respectively).

Conclusion

BMI did not influence the incidence of acute pancreatitis or acute cholangitis on admission, ASA scores, cannulation success or time, post-ERCP complications, length of hospital stays, or 30-day readmission rates. Interestingly, higher CRP and total bilirubin levels were observed in patients without obesity at admission, warranting further investigation into their clinical implications.

References

1. Tokmak S TS. History of ERCP. The Turkish Journal of Academic Gastroenterology 2021;20: 1-2.
2. Duzenli T, Koseoglu H. Endoscopic Retrograde Cholangiopancreatography During the COVID-19 Pandemic: Effects of Enhanced Personal Protective Equipment. Dig Dis Sci 2021;66: 1845-1851.
3. Alfaris N, Alqahtani AM, Alamuddin N, Rigas G. Global Impact of Obesity. Gastroenterol Clin North Am2023;52: 277-293.
4. Deenadayalu VP, Blaut U, Watkins JL, Barnett J, Freeman M, Geenen J, et al. Does obesity confer an increased risk and/or more severe course of post-ERCP pancreatitis?: a retrospective, multicenter study. J Clin Gastroenterol2008;42: 1103-1109.
5. Fujisawa T, Kagawa K, Hisatomi K, Kubota K, Sato H, Nakajima A, et al. Obesity with abundant subcutaneous adipose tissue increases the risk of post-ERCP pancreatitis. J Gastroenterol 2016;51: 931-938.