Yanqun Sun

Sepsis remains the leading cause of in-hospital deaths among children, and there is currently a lack of precise early prediction models. This study aimed to develop an interpretable machine learning (IML) model to predict in-hospital mortality in pediatric patients diagnosed with sepsis according to the 2024 Phoenix Sepsis Diagnostic Criteria. This single-center retrospective cohort study analyzed 464,459 children who were hospitalized in the Children's Hospital Affiliated to Nanjing Medical University from January 2018 to December 2023 were screened. Based on the Phoenix Sepsis Score, 2,477 cases of childhood sepsis were finally diagnosed. The features were filtered using three methods the minimum absolute contraction and selection operator (LASSO) regression, the Boruta algorithm, and random forest importance ranking, ultimately identifying 12 features. Eight machine learning algorithms were trained on 70% of the data and evaluated on a 30% test set. The optimal model was screened based on the accuracy of the test set and the area under the receiver operating characteristic curve (AUC). The interpretability of the model is enhanced by leveraging SHapley Additive exPlanations (SHAP) summary graphs, individual SHAP force maps, and partial dependency plots (PDPs). A total of 2,477 patients with sepsis met the participation criteria, with a median age of 26 months (IQR, 5-78 months). Among the 2,477 participants, 1,448 (58.5%) were boys. The CatBoost model demonstrated the best performance among the 8 constructed models, achieving an AUC of 0.889 and an accuracy of 92.9% in the test set. Notably, at the default decision threshold, the model demonstrated high specificity but relatively low sensitivity. However, by optimizing the decision threshold using the Youden Index, the model's sensitivity was substantially improved to 83.05%, effectively mitigating the potential risk of missed diagnoses for high-risk patients in clinical practice. Feature importance analysis indicated that invasive mechanical ventilation, Glasgow Coma Scale score, platelet count, uric acid, and SpO₂/FiO₂ ratio were the top five features that had the greatest impact on the CatBoost model. SHAP analysis provided both global feature importance and individualised risk explanations, and the model has been deployed as a freely accessible web calculator. The IML model developed in this study provides a highly interpretable and accurate tool for the early prediction of in-hospital mortality in pediatric sepsis patients. The SHAP approach improves model interpretability and helps clinicians understand the factors driving those predictions. However, as a single-center retrospective study, the generalizability of our findings to other clinical settings may be limited. Future multi-center prospective studies are necessary to further validate the model and address its current limitations in sensitivity to minimize the risk of missing high-risk cases.

The coronavirus disease 2019 (COVID-19) pandemic profoundly disrupted the global epidemiology of respiratory syncytial virus (RSV), leading to atypical off-season surges and altering infection dynamics across different climatic zones. This review synthesizes evidence on the landscape of pediatric RSV co-infections in this transformed post-pandemic context. RSV co-infection is frequent, with human rhinovirus (HRV) being the most common viral co-pathogen and Streptococcus pneumoniae (Spn) and Haemophilus influenzae (Hi) predominating as bacterial co-infections. Critically, these co-infections are significantly associated with heightened disease severity, including more intense clinical presentations, prolonged hospitalizations, increased intensive care unit (ICU) admission rates, and greater therapeutic complexity. The relaxation of non-pharmaceutical interventions has been linked to a rebound in co-infection rates. While advances in molecular diagnostics have improved detection, and new prophylactics like nirsevimab offer promise, significant challenges remain. These include gaps in understanding pathogenic synergies, inequities in access to novel interventions, and the need for strategies to manage the ongoing evolution of RSV epidemiology. This underscores the necessity for enhanced surveillance, equitable prevention, and targeted research to mitigate the substantial burden of pediatric RSV co-infections.

Objectives: To characterize the epidemiology, seasonal fluctuations, and co-infection networks of pediatric respiratory pathogens in eastern China following the relaxation of COVID-19 non-pharmaceutical interventions (NPIs). Methods: We conducted a retrospective analysis of 7,473 throat swab samples collected from children (≤18 years) presenting with acute respiratory tract infections (ARTIs) at a tertiary hospital in Nanjing between January 2024 and May 2025. Samples were tested for 13 common pathogens using a multiplex RT-PCR assay. Pathogen detection rates, temporal trends, and co-infection dynamics were analyzed using Kruskal-Wallis tests, time-series smoothing, and network analysis. Results: The overall pathogen detection rate was 61.8% (4,618/7,473), with no significant sex disparity (p = 0.267). Preschoolers (4–6 years) experienced the highest burden (68.3%), significantly exceeding other age groups (p < 0.008). Rhinovirus (RV, 16.9%), Mycoplasma pneumoniae (12.4%), and Respiratory Syncytial Virus (RSV, 9.0%) were the most prevalent pathogens. Distinct seasonal signatures were observed: RSV peaked in winter (25.3% in March 2025), M. pneumoniae dominated in mid-summer (22.8% in August 2024), and RV maintained high year-round activity. Co-infections occurred in 17.5% of positive samples, with M. pneumoniae-RV and Adenovirus-RV being the most frequent combinations. Network analysis identified RV and M. pneumoniae as central connectors within the intricate co-infection landscape. Conclusions: Pediatric respiratory pathogens have returned with strong, age-specific patterns in post-NPI eastern China. The pronounced “immunity debt” among preschoolers and the high rate of co-infections underscore the urgent need for enhanced multiplex surveillance and the expansion of vaccination programs for RSV and influenza.

Background: Communicable diseases (CDs) is the leading cause of death among neonates worldwide, and achieving the Sustainable Development Goal (SDG) 3.2 target (ending preventable neonatal deaths) relies on evidence-based analysis of CDs burden trends and risk factors. This study aims to analyze the temporal trends of CDs burden among neonates in China and the Asia-Pacific region from 1990 to 2021 [based on the Global Burden of Disease (GBD) 2021 database], and project future to 2035, to provide targeted evidence for neonatal health interventions. Methods: This population-based secondary analysis of the GBD 2021 database quantified the incidence, prevalence, mortality, and disability-adjusted life years (DALYs) of neonatal CDs (population denominator: neonates in China and the Asia-Pacific region, 1990–2021). Joinpoint regression analysis was employed to evaluate the temporal trends of disease burden and calculate the estimated annual percentage change (EAPC). The autoregressive integrated moving average (ARIMA) model was adopted to forecast the neonatal CD burden from 2022 to 2035. Age-standardized rates [age-standardized incidence rate (ASIR), age-standardized prevalence rate (ASPR), age-standardized mortality rate (ASMR), age-standardized DALYs rate (ASDR)] were calculated to eliminate the impact of age structure differences on cross-regional comparisons. Results: In 2021, the numbers of incident cases, prevalent cases, deaths cases and DALYs cases of neonatal CDs were 3,741,300.85, 324,827.41, 10,443.43, 943,933.69 in China and 20,407,894.81, 1,786,458.55, 628,615.13, 56,613,320.39 in the Asia-Pacific region. From 1990 to 2021, the age-standardized rates of neonatal CDs showed a significant downward trend in both regions. The EAPC of the ASIR, ASPR, ASMR and ASDR of CDs among neonates was -2.48% [95% confidence interval (CI): −2.69, −2.27], −1.81 (95% CI: −1.89, −1.74), −9.40 (95% CI: −10.09, −8.70), −9.39 (95% CI: −10.08, −8.69) in China and −1.31 (95% CI: −1.40, −1.22), −1.02 (95% CI: −1.14, −0.89), −3.71 (95% CI: −3.88, −3.55), −3.71 (95% CI: −3.88, −3.55) in the Asia-Pacific region, with a higher reduction amplitude in China. Male neonates showed a higher disease burden, though the gender gap gradually narrowed. The ARIMA model predicted a continued decline in neonatal CD burden in China and the Asia-Pacific region by 2035, supporting progress toward SDG 3.2. Low birth weight and short gestation were identified as the leading risk factors, highlighting the critical need to strengthen perinatal care. Conclusions: The study reports a sustained decline in neonatal CDs rates across China and the Asia-Pacific region from 1990 to 2021, marking significant progress towards SDG 3.2. To further reduce disease burden and promote health equity, four targeted interventions are recommended: (I) improving perinatal care to lower preterm birth and low birth weight; (II) strengthening environmental controls and public health infrastructure; (III) implementing gender-specific interventions to address disparities; and (IV) ensuring balanced medical resource allocation, particularly in low- and middle-income countries. Projections indicate a continued decline through 2035, supporting regional and global neonatal health planning.

Background/Objectives: Acute respiratory failure (ARF) is associated with a high mortality. This study aimed to explore the association of arterial partial pressure of carbon dioxide (PaCO₂) in relation to survival outcomes in mechanically ventilated patients with ARF. Methods: This multicenter retrospective cohort study integrated the data from the eICU Collaborative Research Database (eICU-CRD; n = 10,946), the Medical Information Mart for Intensive Care IV (MIMIC-IV; n = 6683), and clinical records from two university-affiliated intensive care units in China (n = 410). The patients were categorized into low, normal, and high PaCO₂ groups using a restricted cubic spline model to explore the relationship between PaCO₂ and mortality. The 28-day survival distributions among the three groups were compared using Kaplan–Meier curves, with statistical significance assessed via the log-rank test. A multivariable Cox proportional hazards model was constructed to evaluate the independent prognostic value of PaCO₂ for multiple complications. Hazard ratios (HRs) and 95% confidence intervals (CIs) were calculated for the low and high PaCO₂ groups relative to the normal PaCO₂ group. Results: A U-shaped relationship was observed between PaCO₂ and mortality, with both low PaCO₂ (57.9 mmHg) associated with an increased mortality risk. Kaplan–Meier survival analysis demonstrated that patients in the intermediate PaCO₂ range (36.4–57.9 mmHg) exhibited the highest survival rate (65.2%), whereas those in the low and high PaCO₂ groups had significantly lower survival rates (60.0% and 63.2%) (log-rank test, p < 0.001). Adjusted survival analyses further revealed that complications such as sepsis and chronic kidney disease significantly influenced the mortality across PaCO₂ strata. Compared with the intermediate PaCO₂ group, the hazard of death increased by 25.5% in the low PaCO₂ group and by 18.9% in the high PaCO₂ group. Conclusions: This retrospective analysis indicates that arterial PaCO₂ levels within the optimal range are associated with improved survival in patients with acute respiratory failure (ARF) on mechanical ventilation, but prospective studies are needed to establish causality and consider potential confounding factors.

This study aims to systematically evaluate the safety of rituximab (RTX) in pediatric nephrotic syndrome (NS) using real-world data from the US Food and Drug Administration Adverse Event Reporting System (FAERS). We analyzed FAERS data from Q1 2013 to Q4 2024, focusing on individual case safety reports (ICSRs) involving NS patients under 18 years old where RTX was the primary suspected drug. Adverse events (AEs) were coded using preferred terms (PTs) and mapped to system organ classes (SOCs). Four pharmacovigilance data analysis methods—reporting odds ratio (ROR), proportional reporting ratio (PRR), multinomial gamma Poisson shrinkage (MGPS), and Bayesian confidence propagation neural network (BCPNN)—were employed to assess drug-related AE risks. Among 275 RTX-related ICSRs, the most frequently reported AEs were hypogammaglobulinemia and granulocytopenia. The strongest signals were observed for CD19⁺ lymphocyte depletion, decreased blood immunoglobulin A (IgA), and decreased blood immunoglobulin G (IgG). The median time to AE onset was 7 days (IQR 0–148.5 days), with the majority of AEs (n = 74) occurring within the first month of RTX treatment. RTX use in pediatric NS is linked to significant immune and hematologic AEs, particularly within the first month of treatment. These findings support tailored monitoring strategies to mitigate risks while leveraging RTX’s therapeutic benefits.

Serratia marcescens (S. marcescens) is a gram-negative opportunistic pathogen causing healthcare-associated infections in neonatal intensive care

To develop and evaluate machine learning models combined with survival analysis for predicting 7-, 14-, and 28-day mortality in critically ill children with acute kidney injury (AKI), identifying key predictors to guide risk stratification and early intervention. Using the Pediatric Intensive Care (PIC) database, we analyzed data from 3,624 children with AKI admitted between 2010 and 2018. Nine machine learning algorithms, including CatBoost, were trained to predict mortality, with feature importance assessed via SHapley Additive exPlanations (SHAP). Time-to-event analyses, including Kaplan-Meier and restricted cubic spline methods, examined the temporal impact of predictors on 28-day mortality, stratified by age and AKI stage. CatBoost achieved the highest area under the curve (AUC) values: 0.871 (95% CI: 0.824–0.918) for 7-day, 0.871 (95% CI: 0.829–0.913) for 14-day, and 0.867 (95% CI: 0.829–0.905) for 28-day mortality. Lactate was the top predictor across all models. Time-to-event analyses revealed a linear association between elevated lactate (cut-off: 1.5 mmol/L) and 28-day mortality (p-overall < 0.001), with stronger effects in infants (0–3 years) and AKI stage 1 patients (HR > 1). Machine learning, particularly CatBoost, combined with survival analysis, accurately predicts AKI-related mortality in critically ill children, with lactate as a pivotal marker. These findings support precision risk stratification and early lactate-targeted interventions, though multicenter validation is needed for clinical adoption.