Objective: To investigate the factors influencing intraoperative blood transfusion in patients with hip fractures and to develop a machine learning (ML) model for predicting this risk. Methods: A total of 424 patients with hip fractures who underwent surgical treatment between November 2022 and March 2025 in our hospital were selected. Key feature variables of intraoperative blood transfusion risk were identified using the Boruta algorithm. Four different ML algorithms—support vector machine (SVM), linear discriminant analysis (LDA), mixed discriminant analysis (MDA), and extreme gradient boosting (XGBoost)—were used to develop predictive models for intraoperative blood transfusion risk. The predictive performance of the four ML models were evaluated using accuracy, precision, receiver operating characteristic (ROC) curves, precision-recall curves (PRC), precision-recall gain curves (PRGC), and F1 scores. Shapley additive interpretation (SHAP) was used to interpret the final model. Results: Among the 424 patients, 77(18.2%) received intraoperative blood transfusion. The Boruta algorithm identified albumin (ALB), activated partial thromboplastin time (APTT), types of anesthesia, types of fracture, and hemoglobin (Hb) as key feature variables for predicting intraoperative blood transfusion risk. In model evaluation, the SVM model outperforms the other three models across multiple metrics, including the area under the receiver operating characteristic curve (AUC), recall, recall gain, accuracy, precision, F1 score, and the area under the precision-recall curve (PRC-AUC). The SVM model, interpreted and visualized based on SHAP values, effectively predicted intraoperative blood transfusion risk in patients with hip fracture. A visual online application was developed based on the SVM model (https://pbo-nomogram.shinyapps.io/blood/). Conclusion: Preoperative low ALB and Hb levels, prolonged APTT, general anesthesia, and intertrochanteric fractures are risk factors for intraoperative blood transfusion in hip fracture patients. The risk prediction model for intraoperative blood transfusion constructed based on the SVM algorithm has optimal performance, which provides new ideas and methods for the clinical early identification of hip fracture patients with high transfusion risk and the implementation of targeted interventions.

Objective To explore the relationship between dynamic contrast-enhanced magnetic resonance imaging(DCE-MRI)and clinical pathological features of invasive breast cancer and lymphovascular invasion(LVI).Methods Imaging and clinical pathological data were retrospectively collected from 508 patients with invasive breast cancer who underwent breast DCE-MRI at the First Medical Center of Chinese PLA General Hospital from January 2019 to August 2021.Patients were divided into the LVI-positive(LVI+)group(n=79)and LVI-negative(LVI-)group(n=429)based on postoperative pathological results.Univariate and multivariate logistic regression analyses were used to identify risk factors for LVI.Results Compared with LVI-group,LVI+group had a higher proportion of patients aged<45 years(44.3%vs.27.0%,P=0.002),non-mass-like enhancement(NME)(31.7%vs.17.7%,P=0.004),Ki-67 expression rate(40.0%vs.30.0%,P<0.001),high Ki-67 expression(94.9%vs.78.1%,P=0.001),Luminal B subtype(76.0%vs.60.1%,P=0.008),and positive axillary lymph nodes rate(72.2%vs.31.5%,P<0.001),while the proportion of Luminal A subtype was lower(2.5%vs.21.5%,P<0.001).Univariate and multivariate logistic regression analyses showed that age≥45 years(OR=0.468,95%CI 0.280-0.783,P=0.004)was an independent protective factor for LVI,while NME(OR=1.987,95%CI 1.126-3.444,P=0.016)was an independent risk factor.Compared with Luminal A subtype,patients with Luminal B subtype(OR=10.482,95%CI 3.164-64.923,P=0.001),HER-2 overexpression subtype(OR=11.571,95%CI 2.755-79.341,P=0.003)and triple-negative subtypes(OR=8.433,95%CI 1.985-57.908,P=0.009)had a higher risk of LVI.Conclusions Age≥45 years is an independent protective factor for LVI,while NME is an independent risk factor.Among molecular subtypes,patients with Luminal B,HER-2 overexpression and triple-negative subtypes have a higher risk of LVI compared with the Luminal A subtype.

With the rapid development of generative artificial intelligence(GAI)technologies,their widespread application in academic research and writing is continuously expanding the boundaries of sci-entific inquiry.However,this trend has also raised a series of ethical and regulatory challenges,inclu-ding issues related to authorship,content authenticity,citation accuracy,and accountability.In light of the growing involvement of AI in generating academic content,establishing an open,controllable,and trustworthy ethical governance framework has become a key task for safeguarding research integrity and maintaining trust within the academic community.This expert consensus outlines ethical requirements across key stages of AI-assisted academic writing-including topic selection,data management,citation practices,and authorship attribution.It aims to clarify the boundaries and ethical obligations surrounding AI use in academic writing,ensuring that technological tools enhance efficiency without compromising in-tegrity.The goal is to provide guidance and institutional support for building a responsible and sustainable research ecosystem.

Objective To propose a dynamic early warning model based on machine learning methods and validate its predi-ctive efficacy so as to achieve precise assessment and early warning of mortality risk in patients with traumatic hypothermia.Methods Firstly,a total of 480 patients who met inclusion criteria were retrospectively selected from the eICU database and randomly divided into training and test sets at an 8∶2 ratio.Secondly,physiological parameters were extracted from these patients,and five machine learning algorithms including XGBoost,AdaBoost,LightGBM,logistic regression(LR)and random forest(RF)were employed respectively to develop dynamic mortality risk warning models for traumatic hypothermia patients,utilizing a 1-hour observation window.Thirdly,receiver operating characteristic curves(ROC)were plotted using the test set data and the effects of different warning windows on the model performance were analyzed by calculating the AUC.Finally,the interpretability of the models was analyzed using the SHapley Additive exPlanations(SHAP)algorithm to elucidate the contribution of each feature to predictive performance.Results The optimal warning window for the dynamic warning model constructed using the eICU database was 12 hours,and in case of 12-hour warning window the logistic regression model achieved the highest AUC of 0.935 and showed optimal predictive performance.The results of the interpretability analysis by the SHAP algorithm showed that body temperature was the feature that had the greatest impact on the model results,and its reduction was positively correlated with the increased risk of death.Conclusion The machine learning-based dynamic warning model for mortality risk in traumatic hypothermia patients enables real-time dynamic risk assessment,providing robust support for clinicians to identify the patient's condition changes at an early stage and references for the adjustment of clinical treatment programs.[Chinese Medical Equipment Journal,2025,46(3):9-14]

Recent data suggest that vascular endothelial growth factor receptor inhibitor (VEGFRi) can enhance the anti-tumor activity of the anti-programmed cell death-1 (anti-PD-1) antibody in colorectal cancer (CRC) with microsatellite stability (MSS). However, the comparison between this combination and standard third-line VEGFRi treatment is not performed, and reliable biomarkers are still lacking. We retrospectively enrolled MSS CRC patients receiving anti-PD-1 antibody plus VEGFRi (combination group, n=54) or VEGFRi alone (VEGFRi group, n=32), and their efficacy and safety were evaluated. We additionally examined the immune characteristics of the MSS CRC tumor microenvironment (TME) through single-cell and spatial transcriptomic data, and an MSS CRC immune cell-related signature (MCICRS) that can be used to predict the clinical outcomes of MSS CRC patients receiving immunotherapy was developed and validated in our in-house cohort. Compared with VEGFRi alone, the combination of anti-PD-1 antibody and VEGFRi exhibited a prolonged survival benefit (median progression-free survival: 4.4 vs. 2.0 months, P=0.0024; median overall survival: 10.2 vs. 5.2 months, P=0.0038) and a similar adverse event incidence. Through single-cell and spatial transcriptomic analysis, we determined ten MSS CRC-enriched immune cell types and their spatial distribution, including naive CD4⁺ T, regulatory CD4⁺ T, CD4⁺ Th17, exhausted CD8⁺ T, cytotoxic CD8⁺ T, proliferated CD8⁺ T, natural killer (NK) cells, plasma, and classical and intermediate monocytes. Based on a systemic meta-analysis and ten machine learning algorithms, we obtained MCICRS, an independent risk factor for the prognosis of MSS CRC patients. Further analyses demonstrated that the low-MCICRS group presented a higher immune cell infiltration and immune-related pathway activation, and hence a significant relation with the superior efficacy of pan-cancer immunotherapy. More importantly, the predictive value of MCICRS in MSS CRC patients receiving immunotherapy was also validated with an in-house cohort. Anti-PD-1 antibody combined with VEGFRi presented an improved clinical benefit in MSS CRC with manageable toxicity. MCICRS could serve as a robust and promising tool to predict clinical outcomes for individual MSS CRC patients receiving immunotherapy.
Humans ; Colorectal Neoplasms/drug therapy* ; Male ; Female ; Immunotherapy ; Middle Aged ; Aged ; Tumor Microenvironment/immunology* ; Retrospective Studies ; Microsatellite Instability ; Transcriptome ; Single-Cell Analysis ; Programmed Cell Death 1 Receptor/immunology* ; Gene Expression Profiling ; Immune Checkpoint Inhibitors/therapeutic use* ; Adult ; Receptors, Vascular Endothelial Growth Factor/antagonists & inhibitors*

Background: s/Aims: Non-invasive models stratifying clinically significant portal hypertension (CSPH) are limited. Herein, we developed a new non-invasive model for predicting CSPH in patients with compensated cirrhosis and investigated whether carvedilol can prevent hepatic decompensation in patients with high-risk CSPH stratified using the new model. Methods: Non-invasive risk factors of CSPH were identified via systematic review and meta-analysis of studies involving patients with hepatic venous pressure gradient (HVPG). A new non-invasive model was validated for various performance aspects in three cohorts, i.e., a multicenter HVPG cohort, a follow-up cohort, and a carvediloltreating cohort. Results: In the meta-analysis with six studies (n=819), liver stiffness measurement and platelet count were identified as independent risk factors for CSPH and were used to develop the new “CSPH risk” model. In the HVPG cohort (n=151), the new model accurately predicted CSPH with cutoff values of 0 and –0.68 for ruling in and out CSPH, respectively. In the follow-up cohort (n=1,102), the cumulative incidences of decompensation events significantly differed using the cutoff values of <–0.68 (low-risk), –0.68 to 0 (medium-risk), and >0 (high-risk). In the carvediloltreated cohort, patients with high-risk CSPH treated with carvedilol (n=81) had lower rates of decompensation events than non-selective beta-blockers untreated patients with high-risk CSPH (n=613 before propensity score matching [PSM], n=162 after PSM). Conclusions Treatment with carvedilol significantly reduces the risk of hepatic decompensation in patients with high-risk CSPH stratified by the new model.

BACKGROUND: The available evidence regarding the benefits of percutaneous coronary intervention (PCI) on patients receiving dialysis with coronary artery disease (CAD) is limited and inconsistent. This study aimed to evaluate the association between PCI and clinical outcomes as compared with medical therapy alone in patients undergoing dialysis with CAD in China. METHODS: This multicenter, retrospective study was conducted in 30 tertiary medical centers across 12 provinces in China from January 2015 to June 2021 to include patients on dialysis with CAD. The primary outcome was major adverse cardiovascular events (MACE), defined as a composite of cardiovascular death, non-fatal myocardial infarction, and non-fatal stroke. Secondary outcomes included all-cause death, the individual components of MACE, and Bleeding Academic Research Consortium criteria types 2, 3, or 5 bleeding. Multivariable Cox proportional hazard models were used to assess the association between PCI and outcomes. Inverse probability of treatment weighting (IPTW) and propensity score matching (PSM) were performed to account for potential between-group differences. RESULTS: Of the 1146 patients on dialysis with significant CAD, 821 (71.6%) underwent PCI. After a median follow-up of 23.0 months, PCI was associated with a 43.0% significantly lower risk for MACE (33.9% [ n  = 278] vs . 43.7% [ n  = 142]; adjusted hazards ratio 0.57, 95% confidence interval 0.45-0.71), along with a slightly increased risk for bleeding outcomes that did not reach statistical significance (11.1% vs . 8.3%; adjusted hazards ratio 1.31, 95% confidence interval, 0.82-2.11). Furthermore, PCI was associated with a significant reduction in all-cause and cardiovascular mortalities. Subgroup analysis did not modify the association of PCI with patient outcomes. These primary findings were consistent across IPTW, PSM, and competing risk analyses. CONCLUSION This study indicated that PCI in patients on dialysis with CAD was significantly associated with lower MACE and mortality when comparing with those with medical therapy alone, albeit with a slightly increased risk for bleeding events that did not reach statistical significance.
Humans ; Percutaneous Coronary Intervention/methods* ; Male ; Female ; Coronary Artery Disease/drug therapy* ; Retrospective Studies ; Renal Dialysis/methods* ; Middle Aged ; Aged ; China ; Proportional Hazards Models ; Treatment Outcome

Objective To propose a dynamic early warning model based on machine learning methods and validate its predi-ctive efficacy so as to achieve precise assessment and early warning of mortality risk in patients with traumatic hypothermia.Methods Firstly,a total of 480 patients who met inclusion criteria were retrospectively selected from the eICU database and randomly divided into training and test sets at an 8∶2 ratio.Secondly,physiological parameters were extracted from these patients,and five machine learning algorithms including XGBoost,AdaBoost,LightGBM,logistic regression(LR)and random forest(RF)were employed respectively to develop dynamic mortality risk warning models for traumatic hypothermia patients,utilizing a 1-hour observation window.Thirdly,receiver operating characteristic curves(ROC)were plotted using the test set data and the effects of different warning windows on the model performance were analyzed by calculating the AUC.Finally,the interpretability of the models was analyzed using the SHapley Additive exPlanations(SHAP)algorithm to elucidate the contribution of each feature to predictive performance.Results The optimal warning window for the dynamic warning model constructed using the eICU database was 12 hours,and in case of 12-hour warning window the logistic regression model achieved the highest AUC of 0.935 and showed optimal predictive performance.The results of the interpretability analysis by the SHAP algorithm showed that body temperature was the feature that had the greatest impact on the model results,and its reduction was positively correlated with the increased risk of death.Conclusion The machine learning-based dynamic warning model for mortality risk in traumatic hypothermia patients enables real-time dynamic risk assessment,providing robust support for clinicians to identify the patient's condition changes at an early stage and references for the adjustment of clinical treatment programs.[Chinese Medical Equipment Journal,2025,46(3):9-14]

BACKGROUND: Based on the China-VHD database, this study sought to develop and validate a Valvular Heart Disease- specific Age-adjusted Comorbidity Index (VHD-ACI) for predicting mortality risk in patients with VHD. METHODS & RESULTS: The China-VHD study was a nationwide, multi-centre multi-centre cohort study enrolling 13,917 patients with moderate or severe VHD across 46 medical centres in China between April-June 2018. After excluding cases with missing key variables, 11,459 patients were retained for final analysis. The primary endpoint was 2-year all-cause mortality, with 941 deaths (10.0%) observed during follow-up. The VHD-ACI was derived after identifying 13 independent mortality predictors: cardiomyopathy, myocardial infarction, chronic obstructive pulmonary disease, pulmonary artery hypertension, low body weight, anaemia, hypoalbuminaemia, renal insufficiency, moderate/severe hepatic dysfunction, heart failure, cancer, NYHA functional class and age. The index exhibited good discrimination (AUC, 0.79) and calibration (Brier score, 0.062) in the total cohort, outperforming both EuroSCORE II and ACCI (P < 0.001 for comparison). Internal validation through 100 bootstrap iterations yielded a C statistic of 0.694 (95% CI: 0.665-0.723) for 2-year mortality prediction. VHD-ACI scores, as a continuous variable (VHD-ACI score: adjusted HR (95% CI): 1.263 (1.245-1.282), P < 0.001) or categorized using thresholds determined by the Yoden index (VHD-ACI ≥ 9 vs. < 9, adjusted HR (95% CI): 6.216 (5.378-7.184), P < 0.001), were independently associated with mortality. The prognostic performance remained consistent across all VHD subtypes (aortic stenosis, aortic regurgitation, mitral stenosis, mitral regurgitation, tricuspid valve disease, mixed aortic/mitral valve disease and multiple VHD), and clinical subgroups stratified by therapeutic strategy, LVEF status (preserved vs. reduced), disease severity and etiology. CONCLUSION The VHD-ACI is a simple 13-comorbidity algorithm for the prediction of mortality in VHD patients and providing a simple and rapid tool for risk stratification.

Background: s/Aims: Non-invasive models stratifying clinically significant portal hypertension (CSPH) are limited. Herein, we developed a new non-invasive model for predicting CSPH in patients with compensated cirrhosis and investigated whether carvedilol can prevent hepatic decompensation in patients with high-risk CSPH stratified using the new model. Methods: Non-invasive risk factors of CSPH were identified via systematic review and meta-analysis of studies involving patients with hepatic venous pressure gradient (HVPG). A new non-invasive model was validated for various performance aspects in three cohorts, i.e., a multicenter HVPG cohort, a follow-up cohort, and a carvediloltreating cohort. Results: In the meta-analysis with six studies (n=819), liver stiffness measurement and platelet count were identified as independent risk factors for CSPH and were used to develop the new “CSPH risk” model. In the HVPG cohort (n=151), the new model accurately predicted CSPH with cutoff values of 0 and –0.68 for ruling in and out CSPH, respectively. In the follow-up cohort (n=1,102), the cumulative incidences of decompensation events significantly differed using the cutoff values of <–0.68 (low-risk), –0.68 to 0 (medium-risk), and >0 (high-risk). In the carvediloltreated cohort, patients with high-risk CSPH treated with carvedilol (n=81) had lower rates of decompensation events than non-selective beta-blockers untreated patients with high-risk CSPH (n=613 before propensity score matching [PSM], n=162 after PSM). Conclusions Treatment with carvedilol significantly reduces the risk of hepatic decompensation in patients with high-risk CSPH stratified by the new model.