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.

Gastrointestinal (GI) cancers account for approximately one-third of annual cancer-related deaths globally, while outcomes remain poor despite advances in surgery, chemotherapy, radiotherapy, and immunotherapy. Given the challenges of persistent resistance and treatment-related toxicities in current therapies, the pivotal roles of the gut microbiota and fecal microbiota transplantation (FMT) in GI cancer therapy are increasingly recognized. This review aims to explore the potential and mechanisms of FMT as a therapeutic adjuvant in the treatment of GI cancers. FMT may enhance antitumor treatment efficacy and reduce treatment-related toxicity through multiple mechanisms, including enhancing antigen presentation, reshaping the tumor microenvironment, and preserving intestinal barrier function. Preliminary clinical evidence indicates that FMT combined with immune checkpoint inhibitors, chemotherapy, or radiotherapy can improve treatment response rates in some trials and may reverse resistance and alleviate associated intestinal toxicities in selected cases. However, clinical application is hindered by donor microbiota functional heterogeneity, substantial interindividual variability in engraftment, and the absence of validated predictive models. To advance FMT toward precision intervention, we propose a functional screening framework: the Healthy Donor-derived Microbiota Xenograft model as a preclinical functional screening platform and its subsequent clinical application, Xenograft-screened FMT, which links donor-level functional validation with personalized microbiota delivery. By integrating mechanistic insights, emerging preclinical and clinical evidence, and a functional screening framework, this review contributes to advancing FMT from an empirical intervention toward a precision adjuvant strategy and offers insights into future clinical investigation of FMT as a therapeutic approach in GI oncology.

Observational studies have shown that educational attainment is associated with the risk of myopia, but the causality of this relationship is unclear. The aim of the present study was to investigate the causal association between educational attainment and myopia.

Using publicly available data from genome-wide association studies, single nucleotide polymorphisms associated with educational attainment (college/university completion and years of education) were selected as instrumental variables. Causal associations with myopia risk were examined using two-sample Mendelian randomization (MR) analyses. Sensitivity analyses were conducted to assess the robustness of the results in terms of violations of MR assumptions.

The inverse variance–weighted analysis revealed potential causal associations of college/university completion (odds ratio (OR) = 1.102; 95% confidence interval (CI): 1.085–1.119; P < 0.001) and years of education (OR = 1.009; 95% CI: 1.007–1.010; P < 0.001) with myopia risk. MR-Egger and weighted median methods yielded similar results for both educational attainment measures.

MR evidence supports a potential causal association between educational attainment and myopia. This evidence highlights the need for careful management of myopia risk in individuals with higher educational attainment.

Autoimmune pancreatitis (AIP) is a rare immune-mediated form of chronic pancreatitis. It may affect multiple organs, and its heterogeneous clinical manifestations complicate diagnosis and management. Based on the Chinese Guidelines for the Diagnosis and Management of Autoimmune Pancreatitis (Shanghai 2012 Draft), together with the latest domestic and international guidelines and research advances, the present guideline provides 20 recommendations covering four aspects: diagnosis, treatment, follow-up, and prognosis. The aim is to improve the diagnosis and management of AIP in China and ultimately improve patient outcomes.

Antimitochondrial antibody (AMA)-negative primary biliary cholangitis (PBC) is a rare presentation of PBC that comprises 5%–10% of all PBC patients. The pathogenesis of AMA-negative PBC appears to be similar to that of AMA-positive PBC. AMA-negative PBC presents similarly to AMA-positive PBC, with symptoms of cholestasis, fatigue, and pruritus most commonly reported. Defective bicarbonate production, resulting in acidification of bile and bile acids, has been proposed as the primary mechanism of damage to bile ducts and hepatocytes and is reflected in elevations of alkaline phosphatase and aminotransferases. Chronic damage can lead to the development of cirrhosis. The diagnosis is made by the finding of AMA negativity by ELISA or assays of similar sensitivity and a positive PBC-specific antinuclear antibody (ANA; anti-glycoprotein 210 and anti-speckled 100 kDa protein) test. In cases in which anti-glycoprotein 210 and anti-speckled 100 kDa protein assays are also negative, a liver biopsy is required to make the diagnosis after exclusion of other causes of cholestasis by magnetic resonance cholangiopancreatography or endoscopic retrograde cholangiopancreatography. Treatment for AMA-negative PBC is the same as for AMA-positive cases, with ursodeoxycholic acid as the first-line treatment. Current treatment is most effective in early stages, where it slows but does not eliminate progression. Risk stratification by validated tools such as the GLOBE and UK-PBC scores remains useful in AMA-negative PBC.

This narrative review provides a detailed and comprehensive examination of hemoadsorption therapy and its role in treating severe poisoning. First, the global problem of suicidal and nonsuicidal self-injury is described, with regional differences in the types of poisons used noted. Lower- and middle-income countries are disproportionately affected by pesticides compared to high-income countries. Organophosphates often constitute the majority of pesticide poisoning in many of these countries. Next, we review the history of hemoadsorption therapy from its early origins to its current evolution. The key physical and chemical principles underlying extracorporeal therapy and its effectiveness are described. A review of the literature examining the evidence for the efficacy of hemoadsorption therapy in poisoning is presented. Current evidence-based guidelines are summarized, including toxin types, clinical indications, and the extracorporeal therapies recommended. Emerging evidence regarding the use of hemoadsorption therapy for severe organophosphate and calcium channel blocker poisoning is also considered. A care pathway for considering hemoadsorption in poisonings where formal guidelines are lacking is proposed. Both the hemoadsorption strategies used and the potential adverse effects of this therapy are discussed. For this narrative review, the PubMed/Medline was searched from inception to April 30, 2025, using the terms (“hemoperfusion” OR “hemadsorption”) AND (“poisoning”). Clinical trials, randomized controlled trials, and meta-analyses were included, along with additional relevant studies identified through a manual review of references. The role of modern resin bead hemoadsorption therapy for severe poisoning is expanding to include removal of commonly encountered poisons that are protein-bound and have a large volume of distribution. Using a multicycle approach, hemoadsorption therapy has shown improved outcomes for both calcium channel blockers and organophosphate poisoning.

The gut microbiota engages in a complex, bidirectional dialogue with the liver via the gut–liver axis, and its dysbiosis plays a central role in the initiation and progression of various liver diseases. This review comprehensively integrates recent advances in the common features and etiology-specific patterns of gut microbial dysbiosis in liver diseases, signal decoding of key microbial metabolite axes, gut–liver immune crosstalk mechanisms, the accelerating role of gut barrier disruption, and recent progress in the use of the microbiome as diagnostic and prognostic biomarkers. We focus on analyzing the common patterns of reduced diversity, depletion of beneficial bacteria, and enrichment of pathogenic bacteria associated with gut flora dysbiosis across different liver diseases, ranging from nonalcoholic fatty liver disease and alcoholic liver disease to cirrhosis and hepatocellular carcinoma, as well as their unique etiology-related characteristics. Core findings reveal that microbial metabolites act as key chemical messengers that precisely drive liver disease progression by modulating host metabolic, immune, and inflammatory pathways. Meanwhile, the translocation of microbes and their products resulting from disruption of gut barrier integrity serves as a key accelerator, exacerbating liver injury and related complications. Based on these mechanisms, this review further explores ecological niche remodeling strategies targeting the gut microbiota, including the existing evidence and limitations of fecal microbiota transplantation and probiotics/prebiotics, as well as the prospects of emerging precision interventions such as phage therapy, microbial enzyme inhibitors, and engineered bacteria. Finally, we emphasize the potential and personalized implementation pathways of synergistically integrating microbiota modulation with existing therapies such as antivirals, antifibrotics, immunotherapy, and metabolic surgery. Future research must focus on promoting the translation of microbiome research from association studies to clinical applications through multi-omics integration and prospective clinical trials, ultimately achieving precise prevention and treatment of liver diseases based on gut–liver axis regulation.

Metabolic dysfunction–associated steatotic liver disease (MASLD) is increasing rapidly, yet regional differences in burden and care quality remain unclear. This study aimed to compare regional incidence, mortality, and disability; evaluate care quality; identify key determinants; and project future incidence.

We analyzed the Global Burden of Disease 2023 estimates of MASLD incidence, deaths, and disability-adjusted life years from 1990 to 2023 by age, sex, country, and region. Age-standardized rates were assessed using joinpoint regression. A composite Quality of Care Index (QCI) was derived through principal component analysis. Gradient boosting models with SHapley Additive exPlanations interpretation identified key predictors, and Bayesian age–period–cohort models generated incidence projections.

In 2023, South and East Asia had the largest numbers of new cases, while North Africa and the Middle East and Andean Latin America recorded the highest age-standardized incidence, mortality, and disability rates. Eastern Europe and Andean Latin America showed sustained increases in mortality and disability despite moderate incidence growth. QCI values were lowest in South Asia, Western Sub-Saharan Africa, and Eastern Europe. High body mass index and fasting plasma glucose were prominent contributors in comparative risk attribution analyses, and machine learning models identified age and calendar year as the strongest predictors of modeled burden patterns. Incidence is projected to continue increasing through 2050, particularly in India and China.

MASLD burden and care quality vary widely across regions. Low-QCI regions show higher mortality and disability, unfavorable metabolic risk profiles, and delayed detection patterns. Strengthening prevention, early case finding, fibrosis assessment, and treatment access may slow MASLD progression.

Pulmonary arterial hypertension (PAH) is a progressive cardiovascular disease with an increasing global burden. Although hemolytic disorders are established causes of PAH, their contribution to the global PAH burden remains unclear. This study aimed to evaluate the association between hemolysis-associated disorders and PAH incidence and to identify the relative contribution of hemolytic disorder subtypes compared with socio-demographic factors.

Using Global Burden of Disease 2021 data, temporal trends in the age-standardized incidence rate (ASIR) of PAH were analyzed using Joinpoint regression. Pearson correlation analysis assessed associations between PAH ASIR and the age-standardized prevalence rates of hemolytic disorder subtypes, hemolysis-related infections, malnutrition, and the Socio-demographic Index (SDI). Random forest regression was used to quantify the contributions of hemolytic disorders to PAH ASIR. Geographic distributions of PAH incidence and hemolytic disorder prevalence were compared, and Bayesian age-period-cohort modeling was used to project their burdens through 2050.

Global PAH ASIR increased from 0.50 to 0.52 per 100,000 from 1990 to 2021. The prevalence of hemoglobinopathies and hemolytic anemias correlated positively with PAH ASIR (R = 0.61, P = 7.70 × 10-22). The random forest model explained 73% of the variance in PAH ASIR (R² = 0.73, P = 0.01), with G6PD trait (percentage increase in mean squared error [%IncMSE]: 18.43), other hemoglobinopathies/hemolytic anemias of unknown etiology (%IncMSE: 18.38), and vitamin A deficiency (%IncMSE: 17.27) identified as the top predictors, surpassing SDI (%IncMSE: 13.25) and sex (%IncMSE: 1.25). Temporal changes in hemolytic disorder prevalence strongly mirrored changes in PAH incidence (R = 0.76, P = 6.34 × 10-39). Exploratory analyses suggested that natural product exposures may contribute to the unexplained hemolytic burden that drives PAH. Projections indicated a continued rise through 2050 in both PAH burden (ASIR increasing from 0.52 in 2022 to 0.57 per 100,000) and hemolytic disease burden (prevalence rising from 27,760.54 to 31,863.72 per 100,000).

Hemolysis-associated disorders, particularly G6PD trait, other hemoglobinopathies/hemolytic anemias, and vitamin A deficiency, are the predominant contributors to the global PAH burden. The projected continued rise in hemolytic disorder prevalence through 2050 signals a persistent exacerbation of the global PAH burden, underscoring the urgent need for targeted prevention strategies.

The aberrant activation of the mTOR pathway and its crosstalk with other signaling cascades represent key drivers of hepatocellular carcinoma (HCC) progression. mTOR-mediated ferroptosis suppression has been implicated in HCC resistance to chemotherapy. This study aimed to elucidate the mechanisms underlying mTOR inhibitor resistance and to evaluate the therapeutic potential of multidrug combinations in β-catenin-mutant HCC.

MHCC97H and SNU449 cells were transfected with 4EBP1WT, 4EBP1A4, or HSP90β expression plasmids and then treated with rapamycin to assess their effects on ferroptosis and rapamycin sensitivity. The role of 4EBP1 in regulating ferroptosis was further explored by Western blotting, co-immunoprecipitation, and immunofluorescence. The inhibitory effects of mTOR inhibitors (rapamycin, MLN0128), ERK inhibitors (PD901), and their combination (MLN0128 + PD901) on tumor cells were evaluated. HCC mouse models were generated via hydrodynamic tail vein injection of c-Met/β-cateninΔN90 or c-Met/β-cateninΔN90/4EBP1A4 plasmids to evaluate the therapeutic effects of the four treatment regimens.

Rapamycin more potently inhibited mTOR/RPS6 than mTOR/4EBP1 and concurrently induced ferroptosis. 4EBP1A4 promoted ferroptosis and potentiated rapamycin efficacy. Mechanistically, 4EBP1A4 competitively bound HSP90β, displacing Keap1, thereby increasing Keap1–Nrf2 complex formation and promoting Nrf2 degradation. Furthermore, rapamycin, MLN0128, PD901, and their combination reduced p-4EBP1 levels, induced ferroptosis, and inhibited HCC cell proliferation, thereby suppressing tumor growth, with the combination exhibiting the strongest effect.

4EBP1A4 enhances Nrf2 ubiquitination and degradation via the HSP90β/Keap1 axis, relieving mTOR-mediated ferroptosis suppression and synergistically improving rapamycin efficacy. Additionally, rapamycin, MLN0128, and PD901 suppress HCC progression by inducing ferroptosis, with their combination showing superior potency.