Uneven distribution of high-quality nephrology care in China has driven rising intercity patient mobility for chronic kidney disease (CKD). This study examined the spatial correlates of this mobility using over 4 million cross-city hospitalization records from 2014 to 2018. First, the Geodetector model was used to identify the key factors and their complex interactions driving patient inflows and outflows, including socioeconomic status, healthcare resource availability, and transportation accessibility. Then, multiscale geographically weighted regression (MGWR) was applied to explore geographical heterogeneity in the influence of these factors on intercity patient mobility. According to the Geodetector model, the leading correlates of patient outflows included hospital bed density, doctor density, and population growth rate, with evident nonlinear and synergistic associations. Patient inflows were mainly influenced by nephrology workforce availability and population structure. MGWR analysis revealed substantial spatial variation in the associations of general and nephrology-specific healthcare resources on intercity patient mobility, underscoring the complex interaction between healthcare capacity and geographic context. This study proposes a novel framework for understanding the spatial correlates of intercity CKD patient mobility in China and highlights the geographic heterogeneity of their associations. The findings support policies aimed at improving the equity and efficiency of CKD care across regions.

Robust evidence is needed for the exposure-response function (ERF) linking chronic kidney disease (CKD) to long-term exposure to ambient fine particulate matter (PM2.5). A three-stage mixed-effects meta-analysis framework was applied to estimate a nonlinear ERF that links PM2.5 exposure to CKD risk ratio or estimated glomerular filtration rate (eGFR) reduction, by integrating all available evidence via systematic searches. Based on the derived ERF, we quantified the global and national burden of CKD attributable to PM2.5 exposure above 10 μg/m3 from 1990 to 2020. A total of 38 studies on CKD and 16 on eGFR were included. Per 10 μg/m3 increment in PM2.5, the pooled risk ratio of CKD was 1.32 (95% confidence interval [CI]: 1.21-1.43) and the ERF indicated a predominantly log-linear association; the pooled eGFR reduction was 2.115 mL/min/1.73 m2 (95% CI: 0.452-3.777), and the ERF showed a saturated effect of high-concentration exposure. In 2020, PM2.5 exposure was estimated to contribute to 485,000 (95% CI: 370,000-587,000) deaths, 5.77 million (95% CI: 4.51-6.99 million) incident cases, and 243 million (95% CI: 188-294 million) prevalent cases of CKD worldwide. Compared with level-2 risk factors for CKD mortality in the Global Burden of Disease study, PM2.5 ranked first in 1990 and second in 2020, reaching a level of importance comparable to major metabolic risk factors. This study provides robust evidence that PM2.5 is a substantial contributor to the global CKD burden and that clean air action is an urgent and integral CKD prevention.

Sepsis has heterogeneous clinical trajectories, but conventional severity scores offer only static risk estimates. Timely, dynamic prediction could enable personalized intervention. In this multicenter retrospective study of 47,936 ICU patients meeting Sepsis-3 criteria from one institutional and two public datasets (MIMIC-III, eICU; sensitivity in MIMIC-IV), group-based trajectory modeling identified latent recovery patterns. An ensemble machine-learning model incorporating dynamic physiological variability was trained, temporally validated, and externally tested; clinical impact was assessed following implementation. Three trajectories emerged: rapid recovery (41.5%), slow recovery (36.4%), and clinical deterioration (22.1%). In the final binary classification task, AUROC was 0.92 (development), 0.89 (internal), 0.84 (MIMIC-III) and 0.77 (eICU); median warning time before deterioration was 17.6 h (Overall pooled across all cohorts). Reduced heart rate variability (SD < 10 bpm) predicted mortality (adjusted HR 2.17). Implementation reduced ICU stay by 1.8 days, machanical ventilation by 2.3 days, and 28-day mortality by 5.7%. This externally validated trajectory-based model offers accurate, early risk stratification for sepsis, supporting proactive, individualized critical care.

In the Anthropocene, high-impact weather (HIW) events are increasingly compressing massive air pollution loads into narrow temporal windows, posing acute health risks that conventional environmental health frameworks were not designed to capture. Despite growing recognition of these dangers, critical gaps persist in the nowcasting of atmospheric health risks, as existing tools such as the air quality index (AQI) and the global burden of disease (GBD) framework remain anchored in chronic exposure paradigms and long-term concentration averages. This manuscript focuses on a pressing standard-index paradox in China: While air quality guidelines/standards (AQG/AQS) are being tightened, the risk-communication AQIs have largely remained static, creating a widening gap in public health protection. As a pilot study, leveraging a recently developed global function between short-term PM 2.5 and cardiovascular mortality (CVD), we conducted a trade-off analysis across AQI cutoff values and identified that the lower boundary for moderately polluted events yields the highest leveraging efficiency. Under the prevailing ambient AQS, the optimal warning threshold is approximately 90 μg/m³, suggesting that health-protective warnings should be extended to cover some slightly polluted days as well. These findings underscore that nowcasting the health risks of acute air pollution episodes is of substantial public health importance, and that a risk-based warning strategy should be considered to replace China’s current AQI-based communication framework.

Fine particulate matter (PM2.5) pollution is a leading cause of the global mortality burden, but public health risks from short-term fluctuations, in addition to long-term exposure, have been insufficiently studied. Potentially influenced by climate change, the daily variability of PM2.5 has significantly increased over the past 60 years, particularly in densely populated regions of East Asia, South Asia, and western North America. To quantify the short-term mortality burden exacerbated by this growing variability, we developed a novel method, the Meta-Regression of Distributional Derivatives (MR-ODDs), to construct nonlinear exposure-response relationships from published summary statistics without requiring individual-level data. Using this approach, we estimate that an average of 672,867 (95% CI: 579,452–740,680) deaths per year globally were attributable to short-term PM2.5 exposure between 2000 and 2023. The mortality burden is especially pronounced in regions where intensifying climate variability drives more frequent extreme pollution events. Western North America, in particular, has emerged as the region with the fastest-rising risk from such events. These findings indicate that extreme pollution events pose a new and growing threat to global public health, demanding a shift in policy from managing annual average concentrations to focusing on early warning and intervention for extreme pollution peaks.

Current approaches to modeling human health are typically confined to a single scale: clinical prediction models focus on individual-level risk estimation from electronic health records, whereas epidemiological models characterize population-level disease dynamics using aggregate or agent-based representations. These paradigms lack a unified framework to capture the bidirectional feedback between individual health trajectories and the evolving environmental, social, and healthcare contexts in which they are embedded. Here, we propose a Multi-Scale Health World Model (MSHWM) that formulates human health as a coupled dynamical system across individual and population scales. The framework consists of two interconnected components: a Micro World Model that learns continuous-time stochastic dynamics of individual health trajectories from longitudinal multimodal data, and a Macro World Model that describes the evolution of population-environment states under both exogenous drivers and endogenous population-level signals. Crucially, cross-scale coupling is established through a distribution-aware interface, in which the empirical population distribution of individual latent states is mapped to a permutation-invariant representation that drives macro-level dynamics, while environmental states directly modulate individual-level transitions. This formulation yields a closed-loop system in which individual trajectories evolve under environment-dependent stochastic dynamics, and the environment co-evolves in response to collective behaviors. By grounding the architecture in a coupled dynamical systems perspective, the proposed framework enables coherent multi-scale simulation, counterfactual reasoning, and long-horizon policy evaluation within a unified generative paradigm.

Per- and polyfluoroalkyl substances (PFAS) persistently challenge conventional water treatment technologies, particularly for short-chain species. This study develops an innovative electro-nanofiltration (E-NF) process that synergistically couples a direct current electric field with nanofiltration to enhance PFAS removal. Systematic investigation reveals that a forward-aligned electric field reduces PFOA permeation flux by 75.2 % compared to conventional NF, primarily via electrophoretic migration. Under optimal conditions (11.1-13.3 V·cm⁻¹), the process achieves high rejection of PFOA (90.4 %) and PFBS (83.9 %) with competitive energy consumption below 1.92 kWh·m⁻³ under optimal conditions. We identify a critical hydraulic threshold at 0.8 MPa, beyond which concentration polarization dominates and leads to chain length-dependent rejection collapse. A modified solution-diffusion-electromigration model quantitatively decouples the transport mechanisms, determining higher apparent electrophoretic mobility for PFBS than for PFOA. The E-NF system demonstrates excellent reversibility and stability, confirming its potential as a robust and energy-efficient solution for mitigating PFAS contamination.

Background: Chronic kidney disease (CKD) is a progressive and clinically heterogeneous condition, contributing to substantial morbidity and premature mortality. However, guideline-directed medical therapy (GDMT) remains insufficiently implemented in CKD care, particularly in under-resourced settings where disease progression is often inadequately monitored and treatment decisions rely largely on episodic clinical encounters. Dynamic, longitudinal decision-support strategies are needed to support timely and individualized treatment optimization. Methods: We developed a deep reinforcement learning (DRL) framework using longitudinal electronic health records (EHRs) to model sequential treatment policies for patients with CKD. This study included 3,869 adults with CKD, identified by ICD-10 codes and at least two post-diagnosis serum creatinine measurements, from the EHR system in Weinan City, China, between 2018 and 2025, contributing 22,876 longitudinal hospital visits. Longitudinal patient trajectories were constructed using demographic characteristics, clinical features, laboratory measurements, and medication histories. A Double Deep Q-Network (DDQN) model was trained to learn treatment policies that maximize cumulative clinical rewards based on patients' longitudinal health states. Model performance was evaluated using five-fold cross-validation. We quantified the GDMT concordance between DDQN-derived treatment recommendations and physician prescriptions based on the KDIGO 2024 guidelines and examined the associations of physician-DDQN prescribing concordance with short-term kidney outcomes and long-term clinical events. Results: The DDQN-derived policy demonstrated substantially greater GDMT concordance than physician prescriptions (56.16% vs. 21.72%). Higher physician-DDQN concordance was consistently associated with more favorable short-term and long-term clinical outcomes. The proportion of patients achieving an eGFR greater than 60 mL/min/1.73 m 2 increased from 61.36% in the lowest concordance group to 90.26% in the highest concordance group. Long-term adverse outcomes also decreased with increasing concordance: kidney failure declined from 23.5% to 4.8%, and all-cause mortality declined from 4.96% to 2.66%. Similar risk reductions were observed for heart failure, cardiovascular disease, and stroke. Conclusion: This study presents a scalable DRL-based approach for dynamic optimization of medical therapy policies in CKD care, with demonstrated potential to improve GDMT adherence and clinical outcomes. These findings support the broader application of data-driven decision-support systems for individualized CKD management, particularly in resourcelimited settings.

Importance: Proactive health has emerged as a transformative paradigm in modern public health, shifting the traditional emphasis from episodic, reactive care toward continuous, anticipatory, and preventive health management. This shift is both timely and necessary in the context of rising chronic disease burdens, aging populations, and increasing demands on healthcare systems. Understanding how advances in medical data and artificial intelligence (AI) underpin this transition is essential for guiding future research and practice. Highlights: This review synthesizes the evolution of proactive health alongside major developments in medical data regimes and AI technologies. The progression of medical AI is characterized across 4 data regimes, including sparse-data, small-data, big-data, and the emerging full-data era characterized by large-scale multimodal data integration. In tandem, AI methodologies have advanced from early expert systems built on rule-based knowledge engineering to contemporary large multimodal models capable of unifying diverse healthcare data streams. These technological and data-centric breakthroughs are reshaping proactive health practices across multiple domains, from decentralized health monitoring, risk-adaptive screening, dynamic and responsive treatment, virtual rehabilitation and digital health intervention, to construction of integrated medical and health services. Conclusions: Despite these advancements, marked technical and ethical challenges remain, limiting the translation of proactive health into routine clinical practice. Addressing these issues will be essential for realizing the full potential of proactive health and enabling future public health systems that are more anticipatory, efficient, and equitable.

IgA nephropathy (IgAN) and IgA vasculitis nephritis (IgAVN) are among the most common glomerulopathies in adolescents, yet treatment options remain limited. Telitacicept (TACI), a dual-target fusion protein, has shown therapeutic potential in adult patients. This study aimed to evaluate the efficacy and safety of telitacicept in adolescents with IgAN and IgAVN. This retrospective observational study included 15 adolescent patients (11 with IgAN and 4 with IgAVN) who received telitacicept between January 2023 and January 2025. Patients received weekly subcutaneous injections of 80 mg or 160 mg. The primary efficacy outcome was proteinuria from baseline to follow-up. Secondary outcomes included changes in eGFR, serum albumin, and hemoglobin. The median age of the patients was 16.91 years. After TACI treatment, the median proteinuria decreased from 2.0 g/day at baseline to 1.0 g/day at month 3 (P = 0.074) and further to 0.5 g/day at month 12 (P = 0.015), corresponding to median reductions of 31.2

This prospective cohort study included 408,760 older adults to investigate complex interaction between waist circumference (WC), blood glucose (BG) or blood pressure (BP), and sex in relation to elderly mortality. We used Cox regression models incorporating a tensor product interaction function to model joint impacts of WC and four cardiometabolic markers on mortality, and developed a two-dimensional exposure-response function (ERF) to quantify the population adaptability to cardiometabolic dysfunction across different WC levels. The linear and nonlinear effects of BG and BP on mortality varied by WC, with significant synergistic interactions. The two-dimensional ERF quantified variations in excess mortality risk across different WC and cardiometabolic marker combinations. Individuals with higher WC exhibited a forward shift in risk thresholds, indicating reduced adaptability to elevated BG and BP levels. Our findings highlight the need for targeted cardiometabolic health management strategies to enhance adaptability and reduce the burden of cardiometabolic diseases in aging populations.

This study aims to establish and validate prediction models based on novel machine learning (ML) algorithms for augmented renal clearance (ARC) in critically ill patients with sepsis. Patients with sepsis were extracted from the Medical Information Mart for Intensive Care IV (MIMICIV) database. Seven ML algorithms were applied for model construction. The Shapley Additive Explanations (SHAP) method was used to explore the significant characteristics. Subgroup analysis was conducted to verify the robustness of the model. A total of 2673 septic patients were included in the analysis, of which 518 patients (19.4%) developed ARC within one week after ICU admission. The Extreme Gradient Boosting (XGBoost) model had the best predictive performance (AUC: 0.841) with the highest balanced accuracy (0.778) and the second-highest NPV (0.950). The maximum creatinine level, maximum blood urea nitrogen level, minimum creatinine level, and history of renal disease were found to be the four most significant parameters through SHAP analysis. The AUCs were higher than 0.75 in predicting ARC through subgroup analysis. The XGBoost ML prediction model might help clinicians to predict the onset of ARC early among septic patients and make timely dose adjustments to avoid therapeutic failure.

The integration of electronic health records (EHRs) into clinical research represents a pivotal advance toward improving data quality, operational efficiency, and regulatory compliance. This review systematically explores recent technological progress in automated data extraction and integration from EHRs, with a focus on three core domains: the development of electronic case report forms (eCRFs), dynamic data exchange mechanisms between EHR and electronic data capture (EDC) systems, and automated standardization strategies to achieve interoperability. eCRFs have evolved from static, manually populated templates to dynamic, technology-enhanced tools that streamline data collection and ensure alignment with clinical protocols and regulatory requirements. Emerging systems increasingly leverage artificial intelligence techniques—including large language models and natural language processing—to automate the generation and population of eCRFs from unstructured EHR data. In parallel, dynamic data pipelines enable real-time, accurate data transfer from EHRs to EDCs, improving data fidelity and reducing clinician workload. Furthermore, the adoption of common data models and standardization frameworks supports multi-center interoperability for large-scale translational research. Despite these advances, widespread implementation is impeded by persistent challenges such as data heterogeneity, governance constraints, and limited scalability. Ongoing efforts to address existing technical challenges are essential optimize the reuse of EHR data in clinical research workflows.

Abstract Objective: Timely access to essential patient information is critical for informed, data-driven decision-making in clinical nursing to enhance care efficiency and quality. Despite advancements in electronic health records, many medical reports remain in paper form, posing challenges for data analysis and application. This study aimed to develop ChatSchema, a pipeline based on Large Multimodal Models (LMMs) for extracting structured information from paper-based medical reports, and evaluated its effectiveness in a pilot setting. Method: ChatSchema was a two-stage approach to extract and structure data from paper-based medical reports. The classification stage leveraged Optical Character Recognition (OCR) of the pictured medical reports, pre-correction, desensitization, and prompt engineering to categorize report types. In the extraction stage, OCR-converted text was transformed into a structured format using a predefined schema, with LMMs applied for standardizing fields and converting data types. A dataset of 100 annotated medical reports was collected from Peking University First Hospital, and the effectiveness of ChatSchema was evaluated in terms of precision, recall, F1-score, and overall accuracy. To validate ChatSchema’s effectiveness, we compared it against a baseline method that provided only schema and task instructions. For sensitivity analysis, two basic LMMs, GPT-4o and Gemini 1.5 Pro, were utilized and compared in the development of ChatSchema. Results: A ground-truth dataset comprising 2,945 test item-result pairs was extracted from 100 annotated medical reports. ChatSchema was capable of correctly extracting and structuring data from paper-based medical reports, showing remarkable accuracy across various data types, unit standardization, field mapping, and basic LMMs. Overall, ChatSchema achieved a high F1-score of 95.8%, an accuracy of 97.2%, a precision of 95.8%, and a recall of 95.8%. ChatSchema surpassed the baseline model by 12.9% in accuracy and by 12.3% in F1-score using the GPT-4o model. Conclusion: Our findings demonstrate that ChatSchema is highly effective for extracting and structuring data from paper-based medical reports. By providing accurate, structured information extraction, ChatSchema has the potential to enhance patient care and support clinical decision-making, showing significant promise for broader application in nursing and healthcare settings.

OBJECTIVE:The impact of desert-originated dust has been underestimated in fine particulate matters (PM2.5)-related disease burden studies. This study aimed to assess the association of long-term dust PM2.5 exposure and all-cause mortality among older adults in China. METHODS:A cohort study using electronic health records (2010-2020) across Weinan, a city in northwest China, which experiences persistently high PM2.5 levels and frequent sand and dust storms, included 1,553,724 adults aged ≥45 years. Annual average dust and non-dust PM2.5 exposures were matched for each participant based on their residential locations. Cox regression models were used to estimate mortality risks associated with long-term exposure to both PM2.5 sources. A tensor product interaction function was applied to develop a two-dimensional joint exposure-response function linking mortality to both dust and non-dust PM2.5 exposures to quantify the burden of deaths attributable to PM2.5 sources. RESULTS:During 8,951,372 person-years of follow-up, 45,273 deaths (2.91 %) occurred. Three-year average exposure to dust and non-dust PM2.5 were 14.85 and 36.78 μg/m3, respectively. Each 1 μg/m3 increase in dust PM2.5 was associated with a 29.0 % (95 % confidence interval [CI]: 28.3, 29.8 %) increase in mortality risk among older adults, versus 13.4 % (95 % CI: 13.1, 13.7 %) for non-dust PM2.5. While dust PM2.5 constituted only 25.9 % of total PM2.5 exposure, it contributed 40.1 % of PM2.5-related deaths in 2019. CONCLUSION:Dust PM2.5 has a stronger adverse association with mortality among older adults than non-dust PM2.5. These findings provide critical insights into PM2.5 source-specific health impacts, informing targeted air pollution control strategies and public health interventions.

Persistent proteinuria is a hallmark of glomerulonephritis and serves as an important diagnostic and prognostic marker. This study proposes a novel digital approach—the quantitative urine protein-to-creatinine ratio (UPCR) home-testing kit (UTK)—for proteinuria screening based on advanced computer vision algorithms and evaluates its effectiveness. Randomly selected spot urine samples were collected from 199 participants recruited from a tertiary hospital in Beijing, China. The UTK method utilized contour detection and image segmentation algorithms to extract the topological information of the urinalysis strip for color calibration, and employed a 3-dimensional color space interpolation algorithm to obtain quantitative UPCR readings. Using the laboratory results as a benchmark, we evaluated the diagnostic performance of UTK in proteinuria screening in terms of validity, reliability, predictive values, and area under the receiver operating characteristic curve (AUC). The mean age of the 199 participants was 51.6 ± 16.2 years and 88 (44.2%) of them were male. The median UPCR was 222.5 mg/g (interquartile range: 106.5–844.6), and 20 (10.1%) participants were identified as having proteinuria. The UTK method performed well in proteinuria screening, with a high accuracy rate of 91.0%, 85.0% sensitivity, a 91.6% specificity, and an AUC of 0.966. The Bland-Altman plot showed a high level of agreement of the quantitative UPCR results between the UTK and laboratory methods. The proposed digital solution for quantitative UPCR analysis based on advanced computer vision technologies showed good performance in proteinuria screening. This user-friendly and cost-effective UPCR measure method presents a promising new strategy for enhancing the efficiency of the primary monitoring and management of glomerulonephritis.