Bacterial infection is a key cause of mortality in patients with acute-on-chronic liver failure (ACLF). In this study, we aimed to identify metabolite biomarkers and develop a novel machine learning model for early identification of bacterial infection in ACLF.

Based on a prospective multicenter cohort from 14 centers, 1,314 patients with acute-on-chronic liver disease were enrolled, including those with ACLF and non-ACLF. Plasma samples at admission were collected for metabolomics profiling. Patients were randomly divided into discovery (n = 921) and validation (n = 393) sets. Machine learning was used to develop diagnostic models. The win ratio method was employed to assess the risk stratification capability of the models.

Bacterial infection occurred in 198 of the 451 ACLF patients and 132 of the 863 non-ACLF patients. Infection altered the plasma metabolome, especially in lipid, amino acid, and xenobiotic metabolic pathways. Models for bacterial infection in ACLF (five metabolites) and non-ACLF (six metabolites) demonstrated superior discrimination in the discovery (AUCs: 0.881 and 0.935, respectively) and validation sets (AUCs: 0.835 and 0.889, respectively) compared with C-reactive protein, white blood cell count, procalcitonin, and the best composite clinical model. Metabolic risk stratification based on the models effectively predicted 90-day outcomes (all-cause death, organ failure, sepsis, new-onset acute decompensation, and systemic inflammatory response syndrome).

Our models based on novel metabolic biomarkers enable identification of patients at high risk of bacterial infection and support risk stratification of 90-day outcomes.

In contemporary practice, elevated 24-hour urinary copper excretion (24-h UCE) often triggers referral for suspected Wilson disease (WD). In this hypercupriuric referral setting, interpretation of 24-h UCE may be distorted by spectrum effects. In this study, we aimed to compare conventional copper biomarkers in hypercupriuric referrals and evaluate whether a Leipzig-aligned ceruloplasmin (Cp) framework could provide a clinically useful triage approach.

We retrospectively studied consecutive, untreated patients evaluated for suspected WD with hypercupriuria between February 2017 and February 2025. The final diagnosis was established using a prespecified Leipzig-based algorithm, with ATP7B testing when indicated. Diagnostic performance of Cp and 24-h UCE was compared. A prespecified Cp three-zone framework was evaluated using <0.10 g/L, 0.10–0.20 g/L, and >0.20 g/L as high-probability, indeterminate, and low-probability zones, respectively.

Among 541 untreated hypercupriuric patients, 65 had WD and 476 had adjudicated non-WD liver disease. Cp outperformed 24-h UCE for diagnosing WD (AUROC, 0.988 vs. 0.762). The optimal Cp cutoff was 0.15 g/L, with 90.8% sensitivity and 97.7% specificity. Cp < 0.10 g/L defined a high-probability zone with 98.0% WD prevalence, whereas Cp > 0.20 g/L defined a low-probability zone with 0.5% WD prevalence. Among non-WD controls, higher urinary copper was independently associated with higher bilirubin, prolonged international normalized ratio, and lower albumin.

In hypercupriuric referrals for suspected WD, Cp retained strong diagnostic performance and outperformed 24-h UCE. A Leipzig-aligned Cp three-zone framework may support probability-based triage in contemporary referral practice.

Primary biliary cholangitis (PBC) significantly impairs health-related quality of life (HRQL), yet the impact of disease stage and fatigue on HRQL and psychological status remains insufficiently quantified. This study aimed to investigate differences in HRQL across disease stages and the impact of fatigue in patients with PBC.

This cross-sectional study recruited 219 patients with PBC from two Chinese tertiary hospitals (2011–2024). After excluding one preclinical case, 218 patients were analyzed. Quality of life was assessed using the validated Chinese versions of the SF-36 and Chronic Liver Disease Questionnaire (CLDQ); psychological status was assessed using the Self-Rating Anxiety Scale and Self-Rating Depression Scale. Between-group differences were quantified by mean differences (MDs) and odds ratios (ORs) with 95% confidence intervals (CIs). Baseline characteristics were balanced across stages (all P > 0.05).

Of the 218 patients (90.4% female; mean age, 57.2 ± 10.3 years), 41 were in the clinical stage, 75 in the fibrosis stage, and 102 in the cirrhosis stage. SF-36 scores were lowest in the cirrhosis stage (e.g., Physical Functioning MD, 17.26; 95% CI, 6.93–27.59 vs. clinical stage), with similar declines in CLDQ domains. Anxiety was highest in the clinical stage (58.5%; OR vs. cirrhosis, 4.13; 95% CI, 1.92–8.92), whereas depression was highest in the cirrhosis stage (55.9%; OR vs. clinical stage, 4.50; 95% CI, 1.95–10.38). Fatigue prevalence was 66.1% and increased with disease stage. Patients with fatigue had lower SF-36 scores in Physical Functioning, Bodily Pain, Vitality, Mental Health, and Physical Component Summary (e.g., Physical Component Summary MD, 38.22; 95% CI, 10.41–66.02).

HRQL declines progressively with PBC stage. Fatigue is strongly associated with impaired HRQL and is closely interrelated with anxiety and depression. Stage-specific psychological patterns suggest the need for tailored supportive interventions.

The modern intensive care unit (ICU) inundates clinicians with large volumes of data, leading to cognitive overload and a gap between data availability and actionable insight. While artificial intelligence (AI) promises a solution, its clinical adoption is limited by systemic barriers, including algorithmic bias, a lack of trust, and validation failures. This paper argues that a design philosophy that envisions AI as an autonomous decision-maker, rather than an integrated collaborative tool, has hindered its clinical adoption. We propose an alternative: a collaborative framework designed to augment the intensivist’s expertise by offloading specific cognitive burdens. This framework redefines AI’s purpose as managing data-intensive tasks, illustrated through four collaborative example roles: a synthesizer to create coherent clinical narratives, a sentinel for proactive deterioration surveillance, a simulator to forecast patient responses to interventions, and a stratifier to identify meaningful subphenotypes within complex syndromes. By delegating these computational tasks, this collaborative model frees clinicians to focus on complex synthesis, nuanced judgment, and compassionate communication. Realizing this vision requires a deliberate translational pathway focused on robust data infrastructure, human-centered design, and rigorous validation through prospective clinical trials. Ultimately, the successful integration of AI in critical care depends not on replacing clinicians but on empowering them, creating a more functional ICU in which technology supports the delivery of safer, more precise, and more humane care.