Yangtze University

Solubility trapping is a key mechanism in geological carbon sequestration (GCS), yet CO₂ solubility in water-filled nanopores often deviates markedly from bulk behavior. We hypothesize that variations in CO₂ solubility within silica nanopores originate from differences in interfacial water structure that are controlled by surface chemistry. In particular, specific Si-OH arrangements on Q2, Q3, and Q4 silica surfaces (defined by the number of Si atoms bonded through oxygen to a central Si atom) modulate hydrogen-bonding (HB) environments and adsorptive volumes that regulate CO₂-water-solid interactions.

We conducted molecular dynamics simulations of water-saturated Q2, Q3, and Q4 silica confinements under representative GCS conditions. Interfacial density profiles, HB distributions, and CO₂ spatial probability maps were analyzed to quantify fluid-solid interactions and to evaluate CO₂ solubility relative to bulk water.

Hydrophilic Q3 surfaces exhibit enhanced CO₂ solubility compared to the bulk liquid, arising from CO₂-water co-adsorption facilitated by a dense interfacial HB network. Hydrophobic Q4 confinements, by contrast, show pronounced over-solubility dominated by strong direct CO₂ adsorption within enlarged low-HB regions. Q2 surfaces display intermediate behavior reflecting mixed hydrophilic-hydrophobic character. We introduce two mechanistic descriptors, adsorptive volume and HB site density. High HB site density promotes hydrophilicity and co-adsorption, whereas large adsorptive volume favors direct CO₂ adsorption and over-solubility. Overall, these results demonstrate that CO₂ solubility in silica nanopores is jointly governed by interfacial water structure and surface chemistry. The findings provide molecular-scale insights into solubility trapping in silica-rich formations and inform the design of engineered materials for CO₂ capture and storage.

Depression and insulin resistance (IR) are highly comorbid. Previous studies, however, have relied on static, single-point IR assessments, failing to capture the prognostic significance of its dynamic changes. This study aims to investigate the association between longitudinal trajectories of the estimated glucose disposal rate (eGDR) and new-onset depression risk.

We analyzed data from adults aged ≥45 years without baseline depression from China Health and Retirement Longitudinal Study (CHARLS). eGDR was calculated from waist circumference, hypertension, and HbA1c. Trajectories were identified using K-means clustering. Depression was defined as a CES-D score ≥ 10. Cox models assessed depression risk. Restricted cubic splines evaluated nonlinear relationships of baseline eGDR and ΔeGDR with depression risk.

Among 4380 participants, three eGDR trajectories were identified: rapid decline (11.5 %), consistently low (25.9 %), and consistently high (62.6 %). Compared to the persistent low trajectory, the rapid decline and persistent high trajectories were associated with a 23.2 % increased (HR = 1.232, 95 % CI: 1.020-1.489) and a 16.2 % decreased (HR = 0.838, 95 % CI: 0.723-0.971) risk of depression, respectively. Both the rapid decline and consistently low trajectories represent higher-risk phenotypes compared to the consistently high trajectory. RCS revealed an inverted U-shaped association between baseline eGDR and depression risk (P < 0.001), with the peak risk located at eGDR = 8.634 (HR = 1.44). Similarly, ΔeGDR showed a U-shaped association with risk; a sharp decline (ΔeGDR ≤ -4.81) was associated with an HR > 1.5.

Longitudinal eGDR trajectories are independent predictors of depression risk. Maintaining high insulin sensitivity is protective, while a rapid decline confers significant risk. Monitoring dynamic eGDR changes is crucial for identifying high-risk individuals and implementing preemptive interventions.

The incidence of refeeding syndrome (RFS) in critically ill patients has attracted increasing attention, accompanied by the proliferation of risk prediction models designed to identify susceptible individuals. Nevertheless, the validity and practicality of these models in clinical settings and for future research are not yet fully established.

The aim of this study was to systematically review published studies on risk prediction models for RFS in critically ill patients.

The following databases were searched from their inception to March 15, 2025: China National Knowledge Infrastructure, Wanfang database, China Science and Technology Journal database (VIP), SinoMed, PubMed, Web of Science, Cochrane Library, CINAHL, and Embase. The Prediction Model Risk of Bias Assessment Tool checklist was used to evaluate the risk of bias and applicability of the included studies.

A total of 1472 studies were retrieved. This review included 24 prediction models from 16 studies after the selection process. The reported area under the receiver operating characteristic curve ranged from 0.626 to 0.980. All studies have a high risk of bias, primarily due to the limited availability of data sources and insufficient reporting of analytical methods.

All studies included in the review on risk prediction models for RFS in critically ill patients were identified as having a high risk of bias. Future studies should prioritise the development of new models with larger sample sizes, rigorous study designs, and multicentre external validation to enhance the reliability and applicability.