Citations

e16017 Background: Pancreatic ductal adenocarcinoma (PDAC) is the most common form of pancreatic cancer and remains one of the deadliest malignancies worldwide, largely due to late-stage diagnosis. Early-stage PDAC is frequently asymptomatic, limiting opportunities for timely detection and intervention. We aimed to develop prognostic models using routinely acquired pre-diagnostic clinical data to identify individuals at high risk for PDAC prior to clinical manifestation. Methods: We conducted a retrospective study of patients treated at Mayo Clinic sites in Rochester, MN; Phoenix, AZ; and Jacksonville, FL between January 2011 and February 2023. The cohort included 1,151 patients with PDAC and 3,745 noncancer controls. CT scans were standardized to axial abdominal acquisitions with slice thickness between 0.8–3.0 mm. Radiology reports were de-identified, normalized, and parsed into structured sections (Indications, Findings, Impression), with an additional pancreas-specific section aggregating pancreas-related sentences. Demographic and clinical variables, e.g., age, sex, race, etc., were encoded using a templated sentence representation. All textual inputs were embedded using a Sentence-BERT model. A multimodal deep learning survival model was developed to integrate CT imaging, radiology report embeddings, and demographic features. The model was trained using a negative log-likelihood loss to estimate patient-specific PDAC risk scores. Model performance was assessed using the concordance index (C-index). Internal validation included patients from MN and AZ sites, while external validation used the FL cohort. Results: We evaluated unimodal (imaging, text, and demographics) and multimodal inputs. Fusion of imaging, radiology report, and demographic features achieved the highest performance, with internal and external validation C-indices of 0.885 (95% CI, 0.856–0.908) and 0.751 (95% CI, 0.719–0.783), respectively. Subgroup analysis by tumor location demonstrated differential performance. For pancreatic head tumors, internal and external C-indices were 0.916 (95% CI, 0.841–0.960) and 0.778 (95% CI, 0.730–0.823). For pancreatic body tumors, C-indices were 0.954 (95% CI, 0.897–0.997) internally and 0.929 (95% CI, 0.897–0.953) externally. For pancreatic tail tumors, internal and external C-indices were 0.903 (95% CI, 0.830–0.974) and 0.789 (95% CI, 0.689–0.874). Conclusions: A multimodal deep survival model leveraging pre-diagnostic CT imaging, radiology report embeddings, and demographic data can effectively identify individuals at high risk for PDAC prior to clinical diagnosis. Differences in performance were observed across pancreatic head, body, and tail tumors. This approach highlights the potential of opportunistic cancer risk stratification from routine imaging to enable earlier detection and targeted intervention strategies.

ABSTRACT Background and Aim Determining the location of gastrointestinal (GI) bleeding is critical for guiding diagnostic and therapeutic strategies but often results in patients receiving pan‐endoscopy with upper endoscopy, colonoscopy, and small bowel capsule endoscopy, which can be unnecessary, costly, and waste limited healthcare resources. The blood urea nitrogen‐to‐creatinine ratio (BUN/Cr) may help discriminate between upper GI bleeding (UGIB) and lower GI bleeding (LGIB). This study aims to assess the diagnostic performance of the BUN/Cr in distinguishing UGIB from LGIB. Methods A literature search was conducted in MEDLINE, EMBASE, and Cochrane Library to identify diagnostic test accuracy studies published from inception through January 31, 2025, which evaluated BUN/Cr in patients with UGIB versus LGIB. A linear mixed‐effects model meta‐analysis was conducted using the “diagmeta” package. The pooled sensitivity, specificity, diagnostic odds ratio, and area under the summary receiver operating characteristic curve (AUC) were estimated, and optimal threshold was determined using the Youden index. Results Seventeen studies were included in this meta‐analysis. The pooled mean difference in BUN/Cr between UGIB and LGIB patients was 11.44 (95% confidence interval [95% CI] 8.35–14.52). At a threshold of > 30, pooled sensitivity was 38.8% (95% CI 28.1%–50.8%) and specificity was 89.3% (95% CI 81.0%–94.2%). At the optimal cut‐off of 22, sensitivity was 66.2% (95% CI 57.8%–73.7%), specificity was 71.0% (95% CI 58.5%–81.0%), and AUC was 0.740 (95% CI 0.683–0.794). Conclusion BUN/Cr demonstrates moderate diagnostic accuracy for distinguishing UGIB from LGIB, particularly at a threshold of ≥ 22. This readily accessible and inexpensive blood test may assist in early clinical triage and assessment of the location of GI bleeding, particularly in those patients where the clinical presentation is obscure.

In the realm of innovative medical procedures, TEER (transcatheter edge-to-edge repair) has emerged as a promising field, showcasing significant growth and advancements. Mitral TEER has been performed for the last two decades; in contrast, tricuspid TEER is newer, with long-term outcomes pending. This article aims to provide a comprehensive review of the current literature, with a primary focus on outcomes and potential complications associated with both procedures. Both procedures carry a low risk of complications when done by experienced providers. A team approach involving specialists in cardiology, cardiothoracic surgery, cardiac imaging and heart failure ensures comprehensive care. A unified approach encompassing preprocedural workup, risk assessment, and standardised care throughout the procedure and recovery contributes to successful outcomes.

Background: cardiovascular diseases, including acute myocardial infarction (AMI) and takotsubo cardiomyopathy (TTC), are significant causes of morbidity and mortality worldwide. Timely differentiation of these conditions is essential for effective patient management and improved outcomes. Methods: We conducted a review focusing on studies that applied artificial intelligence (AI) techniques to differentiate between acute myocardial infarction (AMI) and takotsubo cardiomyopathy (TTC). Inclusion criteria comprised studies utilizing various AI modalities, such as deep learning, ensemble methods, or other machine learning techniques, for discrimination between AMI and TTC. Additionally, studies employing imaging techniques, including echocardiography, cardiac magnetic resonance imaging, and coronary angiography, for cardiac disease diagnosis were considered. Publications included were limited to those available in peer-reviewed journals. Exclusion criteria were applied to studies not relevant to the discrimination between AMI and TTC, lacking detailed methodology or results pertinent to the AI application in cardiac disease diagnosis, not utilizing AI modalities or relying solely on invasive techniques for differentiation between AMI and TTC, and non-English publications. Results: The strengths and limitations of AI-based approaches are critically evaluated, including factors affecting performance, such as reliability and generalizability. The review delves into challenges associated with model interpretability, ethical implications, patient perspectives, and inconsistent image quality due to manual dependency, highlighting the need for further research. Conclusions: This review article highlights the promising advantages of AI technologies in distinguishing AMI from TTC, enabling early diagnosis and personalized treatments. However, extensive validation and real-world implementation are necessary before integrating AI tools into routine clinical practice. It is vital to emphasize that while AI can efficiently assist, it cannot entirely replace physicians. Collaborative efforts among clinicians, researchers, and AI experts are essential to unlock the potential of these transformative technologies fully.