Colorectal polyp detection from endoscopic images is critical for the early diagnosis of colorectal cancer. However, traditional deep learning methods often suffer from limited generalization when deployed across datasets containing different polyp morphologies. This work aimed to investigate whether vision-language foundation models can facilitate zero-shot generalization across multiple polyp datasets without target-domain fine-tuning.

We introduced a zero-shot colorectal polyp detection framework based on Contrastive Language-Image Pretraining (CLIP) to improve cross-dataset detection performance. Key innovations include: (1) a background patch contrastive loss using pseudo-normal tissue patches to teach the model to distinguish normal mucosa from polyps; (2) attribute-enhanced text prompts that incorporate domain-specific descriptors of polyp appearance, improving the model’s semantic generalization to novel polyp morphologies; and (3) an enhanced CLIP visual adapter with per-layer adaptive feature fusion and generalized mean pooling to capture multi-scale features for better polyp localization. During training, we use one annotated colorectal polyp dataset (e.g., CVC-ColonDB) to learn patch-level image-text correspondence. The model is then evaluated in a zero-shot manner on different polyp datasets (CVC-ClinicDB, Kvasir-SEG, and CVC-300), where we measure both pixel-level.

The framework demonstrated robust zero-shot generalization on unseen test cohorts. Without any dataset-specific fine-tuning, the model achieved a mean pixel-level AUROC of 0.94 and a mean average precision of 0.81 across the 12 leave-one-dataset-out zero-shot transfer settings. In the CVC-ColonDB-source benchmark, the model achieved a mean Dice coefficient of 0.84 across CVC-ClinicDB, Kvasir-SEG, and CVC-300. This high level of performance was consistent across datasets with distinct visual characteristics, underscoring the ability of the model to detect diverse polyp morphologies that it had not been explicitly trained to recognize.

Our findings demonstrate that an anomaly-aware vision-language model significantly improves cross-dataset polyp detection generalization without requiring normal images for training. This multimodal strategy may facilitate the robust deployment of artificial intelligence-based colorectal screening systems by enabling reliable detection of diverse polyp morphologies across different clinical settings. Extension to non-polyp colorectal pathologies (e.g., ulcerative colitis and colorectal tumors) remains an important direction for future work, pending the availability of pixel-level annotated datasets for these lesion categories.

Disorders of gut–brain interaction (DGBIs) encompass some of the most common gastrointestinal disorders and affect up to 40% of the general population. Despite their inherent heterogeneity and diverse clinical manifestations, many of the underlying pathophysiological mechanisms overlap among different DGBIs. Activation of the gastrointestinal mucosal immune system at a low level (“low-grade inflammation”) and impairments in gut epithelial barrier structure and function have been reported to play a key role in the pathophysiology of multiple DGBIs, but these alterations cannot be detected using routine clinical testing. Confocal laser endomicroscopy (CLE) is an established, readily available technology that can be added to standard gastrointestinal endoscopy, enabling “real-time” microscopic evaluation of the gastrointestinal surface epithelium. CLE has been found to be capable of identifying gastrointestinal mucosal abnormalities that are reflective of epithelial barrier impairment and/or low-grade immune activation. Over the past several years, multiple intriguing studies have utilized CLE as a clinically applicable tool to evaluate the intestinal mucosa in patients with various DGBIs. The aim of this narrative review is to summarize the available literature on the role of CLE in patients with DGBIs and to provide a perspective on the use of this technology in DGBIs.

Liver fibrosis is a central pathological process driving the progression of chronic liver disease, yet effective antifibrotic therapies remain limited. Increasing evidence has identified the mineralocorticoid receptor (MR), a ligand-activated nuclear receptor, as a key regulator of intrahepatic homeostasis and fibrogenesis. MR is expressed across multiple hepatic cell types, including hepatocytes, hepatic stellate cells, macrophages, and liver sinusoidal endothelial cells, where it integrates metabolic, inflammatory, and microvascular signaling. Under pathological conditions, MR activation—mediated by both aldosterone-dependent and ligand-independent mechanisms such as hypoxia and oxidative stress—amplifies core profibrotic pathways, including TGF-β signaling, reactive oxygen species (ROS) generation, and NF-κB–driven inflammation. These molecular mechanisms are executed in a cell-type–specific manner, promoting hepatic stellate cell activation, macrophage-mediated inflammation, hepatocyte metabolic dysfunction, and liver sinusoidal endothelial cell capillarization, thereby forming a self-reinforcing fibrogenic network. Preclinical studies consistently demonstrate that mineralocorticoid receptor antagonists attenuate fibrosis by targeting these interconnected pathways. However, clinical evidence remains limited, with only early-phase trials in metabolic dysfunction-associated steatohepatitis and indirect support from cardiorenal studies. Nonsteroidal mineralocorticoid receptor antagonists, particularly finerenone, exhibit improved receptor selectivity and safety profiles, highlighting their therapeutic potential. Future research should focus on disease-specific patient stratification, validated antifibrotic endpoints, and rigorous safety evaluation to enable effective clinical translation of MR-targeted therapies in liver fibrosis.

Lipid metabolism reprogramming drives malignant proliferation and invasiveness in hepatocellular carcinoma (HCC). Beyond supplying energy and membrane components, lipids function as signaling molecules that modulate tumor cell epigenetics and the microenvironment. Accumulating research has clarified the implications of these metabolic alterations in HCC, providing a rationale for targeted therapies. This review summarizes key alterations in lipid metabolism within HCC and explores their mechanistic contributions to tumor progression. It further examines how lipid metabolic shifts in immune and stromal cells of the tumor microenvironment promote HCC advancement. Finally, we discuss the therapeutic potential of targeting lipid metabolism in liver cancer treatment.

The Asia-Pacific region accounts for a larger share of the hepatitis B burden than any other regions of the world, presenting a challenge to meeting the World Health Organization (WHO) 2030 elimination targets. In this study, we aimed to quantify the burden of chronic hepatitis B (CHB) and project its trends through 2030 using the GBD 2023 framework, thereby identifying gaps and priorities for the Asia-Pacific region to achieve WHO 2030 targets.

Using data from the Global Burden of Disease Study 2023, we analyzed chronic hepatitis B (CHB) prevalence, mortality, and disability-adjusted life years. We evaluated temporal trends (1990–2023) using average annual percent changes and projected the 2024–2030 burden using Bayesian age-period-cohort models.

In 2023, the Asia-Pacific region accounted for 63% of global CHB cases (178.0 million), 66% of deaths (259.1 thousand), and 65% of disability-adjusted life years (8.4 million). Regional prevalence and mortality rates exceeded global averages, although childhood (<5 years) prevalence was comparatively lower (590.3 vs. 1,325.3 per 100,000). East Asia bore the highest absolute burden (99.2 million cases), and South Asia had the largest pediatric caseload. Between 1990 and 2023, Western Asia showed the steepest decline in adult prevalence (−1.99%), whereas Southeast and Central Asia exhibited upward mortality trends. Projections indicate that the Asia-Pacific region is off track to meet the WHO 2030 disease elimination targets, as the prevalence rate in children under five years remains above the 0.1% target threshold and absolute mortality is projected to increase.

The Asia-Pacific region continues to contribute the largest share of the global CHB burden and now faces persistent gaps despite progress. Although substantial progress has been made in reducing prevalence through immunization, the region is currently off track to meet the WHO 2030 targets for both incidence and mortality.

Porto-sinusoidal vascular disease (PSVD) is a non-cirrhotic vascular liver disorder characterized by portal and sinusoidal microvascular lesions and is frequently complicated by portal hypertension. Accurate assessment of portal pressure is essential for diagnosis, risk stratification, therapeutic decision-making, and prognostic evaluation in PSVD. However, unlike cirrhosis, portal hypertension in PSVD is predominantly presinusoidal, making hepatic venous pressure gradient measurement prone to underestimating true portal pressure. Recent advances have promoted a transition from conventional invasive assessment toward a multimodal and precision-oriented strategy integrating non-invasive and minimally invasive techniques. Ultrasound elastography, computed tomography, and magnetic resonance imaging—particularly radiomics-based approaches—provide valuable tools for differentiating PSVD from cirrhosis and estimating the severity of portal hypertension. Endoscopic ultrasound-guided portal pressure gradient measurement has emerged as a promising minimally invasive technique for direct hemodynamic assessment and prognostic stratification. In addition, laboratory biomarkers, digital modeling, and artificial intelligence-assisted analysis may further improve individualized risk prediction and dynamic monitoring. This review summarizes current advances in portal pressure assessment in PSVD, critically discusses the strengths and limitations of existing approaches, and highlights future directions toward non-invasive, digital, and precision-guided management.

Cardiorenal syndrome is associated with high morbidity and mortality and is characterized by bidirectional interactions between cardiac and renal dysfunction. The advent of sodium-glucose cotransporter 2 inhibitors (SGLT2i) and the nonsteroidal mineralocorticoid receptor antagonist finerenone has substantially changed the therapeutic landscape. Combination therapy with SGLT2i and finerenone may provide additional benefits through complementary mechanisms, representing a potential paradigm shift in the management of cardiorenal syndrome. In this review, we examine the pathophysiological pathways that characterize cardiorenal syndrome, clinical data from major randomized controlled trials, and the rationale for the concomitant use of these two drug classes. SGLT2 inhibitors significantly reduce hospitalization for heart failure, slow renal function decline, and provide benefits in both heart failure with reduced ejection fraction and heart failure with preserved ejection fraction, irrespective of diabetes status. Finerenone has been shown to reduce the risk of cardiovascular events and chronic kidney disease progression in patients with type 2 diabetes and chronic kidney disease, with a more favorable safety profile than steroidal mineralocorticoid receptor antagonists. Emerging evidence suggests that combination therapy may reduce hospitalizations for heart failure and slow renal disease progression beyond the effects of either monotherapy. However, implementation of these therapeutic options requires careful patient selection, ongoing monitoring of renal function and electrolytes, and close collaboration between cardiologists and nephrologists.

Generative artificial intelligence (AI), particularly large language models (LLMs) and multimodal systems, is emerging as a potentially important innovation in intensive care medicine. The intensive care unit (ICU) is a data-dense, high-acuity setting where rapid and accurate decisions are critical. These models can translate complex multimodal data into interpretable and clinically actionable insights across diagnostic, prognostic, and documentation workflows. This review outlines six key domains in which generative AI is currently being explored for its potential to reshape critical care: clinical decision support; clinical documentation automation (AI scribe, voice-to-note); predictive analytics, including sepsis and acute respiratory distress syndrome prediction, acute kidney injury management, ventilator liberation readiness, delirium monitoring, and continuous renal replacement therapy optimization; ICU data summarization and multimodal monitoring; synthetic data generation; and legal and ethical governance. In clinical decision support, hybrid models that integrate time-series monitoring data with LLMs can contextualize alerts, generate diagnostic suggestions, and offer treatment plans with explainable reasoning. Documentation tools that leverage ambient listening and voice-to-note AI can streamline progress notes and discharge summaries, thereby reducing clinician workload. In predictive analytics, LLMs enhance model performance by augmenting sparse electronic health record data and translating outputs into interpretable narratives. Synthetic data generation enables algorithm development and training, particularly for rare events, while protecting patient privacy. However, the realism and ethical deployment of such data require rigorous validation. Widespread implementation of generative AI will require careful attention to challenges related to trust, validation, bias, liability, and regulatory compliance. The use of these tools must remain under clinician supervision to ensure transparency and accountability. With responsible deployment, generative AI may augment ICU workflows, improve outcomes, and reduce clinician burden, potentially becoming an indispensable component of critical care delivery.