Background Long working hours, as a common risk factor for occupational stress, is closely related to the occurrence of depressive symptoms. Understanding how long working hours affect occupational stress and depressive symptoms will inform occupational health interventions. Objective To quantify the impact of long working hours exposure on occupational stress and depressive symptoms among Internet industry employees, translate black-box outputs into actionable insights, and demonstrate the value of interpretable machine learning for early-warning occupational-health surveillance. Methods A dataset was derived from a cross-sectional survey involving 2866 internet industry employees in China. This survey was part of the project Risk Assessment Of Long Working Hour Exposure And Its Adverse Health Effects, conducted by the National Institute for Occupational Health and Poisoning Control, Chinese Center for Disease Control and Prevention, from 2021 to 2023. Working hours, occupational stress and depressive symptoms were quantified with a set of structured questionnaires including the Core Occupational Stress Scale and the Patient Health Questionnaire. Pairwise associations were screened by Mantel tests and variance-inflation factors. Key predictors identified through feature selection were fed into six machine-learning risk-prediction models. Visual interpretation was provided by feature importance, Shapley additive explanations (SHAP) and local interpretable model-agnostic explanations (LIME), while directed causal effects and intervention impacts of prolonged working hours exposure on occupational stress and depressive symptoms were dissected with causal explanation of features techniques. Results The positive rates of occupational stress and depressive symptoms among internet employees were 12.9% and 77.8% respectively. Twelve core features for occupational stress and nine for depressive symptoms were retained after selection. After these features were supplied to six predictive algorithms and evaluated on five metrics, the Light Gradient Boosting Machine (LGBM) achieved the highest accuracy—0.89 for occupational stress and 0.79 for depressive symptoms on the hold-out test set. The feature-importance rankings converged on fatigue accumulation and life satisfaction as dominant drivers for both outcomes, whereas weekly working hours and daily overtime emerged as the principal exposure-related predictors. The SHAP summary plots revealed that longer weekly hours and daily overtime systematically elevated the probability of occupational stress. The causal feature explanation further quantified that ascending one category in weekly working hours increased the probability of occupational stress by 7.04%. Conclusion Exposure to long working hours is associated with both occupational stress and depressive symptoms among internet industry employees. Interpretable machine-learning frameworks translate these associations into transparent, defensible drivers, enabling precise identification of the pivotal factors and their interplay. This evidence base equips occupational-health practitioners with actionable insights for designing targeted prevention and intervention strategies.

OBJECTIVE To analyze pathogens of multiple drug-resistant organism(MDRO)infection and its risk factors during the recovery period of patients with traumatic brain injury,and to construct and validate the predic-tion model for MDRO lung infection.METHODS Patients who were admitted to the Shanxi Second People's Hos-pital for traumatic brain injury(TBI)and developed hospital-acquired pneumonia(HAP)during their hospitaliza-tion between Aug.2019 and Aug.2023 were selected as the study subjects,their clinical data were retrospectively collected,and the patients were randomly divided into a modeling group(n=270)and a verification group(n=90)at a ratio of 3∶1,and they were further categorized into the MDRO infection group(148 cases)and the non-infection group(122 cases)based on whether they developed pneumonia caused by MDRO infections.Clinical data of the patients were collected for analysis,with pathogen identification performed for all patients.The risk factors for MDRO infection in TBI patients with HAP were analyzed by multivariate Logistic regression analysis,a pre-diction model for MDRO infections during the hospitalization and recovery period of TBI patients with HAP was constructed using a formula-based method,the goodness-of-fit of the model was assessed using Hosmer-Leme-show test,and its predictive ability was evaluated using receiver operating characteristic(ROC)curves.RESULTS There was no significant differences in the baseline characteristics of TBI patients with HAP between the modeling group and the verification group.The incidence of MDRO infections among TBI patients with HAP in modeling group was 54.81％(148/270),with Klebsiella pneumoniae(25.27％)being the predominant pathogen.Mechan-ical ventilation duration(OR=5.789,95％CI:2.164-15.486,P=0.001),ICU length of stay(OR=5.441,95％CI:2.153-13.751,P＜0.001),combined kidney diseases(OR=2.770,95％CI:1.321-5.812,P=0.007),duration of antibiotics use(OR=7.486,95％CI:1.928-29.059,P=0.004),occurrence of impaired consciousness(OR=5.720,95％CI:2.586-12.652,P＜0.001)and use of aminoglycoside antibiotics(OR=3.861,95％CI:1.608-9.273,P=0.003)were risk factors for MDRO infections.The Hosmer-Lemeshow for the risk prediction model showed that x2=5.013 and P=0.496.ROC curves analysis showed The area under the curve(AUC)for the modeling group by ROC curve analysis was 0.80(95％CI:0.773-0.877),with a sensitivi-ty of 83.33％and a specificity of 73.60％respectively,and the AUC for the verification group were 0.800(95％CI:0.670-0.895),with a sensitivity of 81.82％and a specificity of 77.27％respectively.CONCLUSION The prediction model constructed in this study,based on the risk factors for MDRO infection pneumonia during the re-covery period in hospitalized patients with traumatic brain injury,demonstrated high goodness-of-fit,as well as satisfactory sensitivity and specificity,and its application in clinical practice may help identify patients at high risk of MDRO infections.

Objective To compare the consistency of voice parameters collected by commonly used smart-phone brands in China and professional recording equipment,and to study whether smartphones can be used for voice research.Methods A total of 67 normal subjects were selected for voice recording using six different smart-phone brands(via the"Active Health"screening APP from the National Key Research and Development Program)and professional recording equipment.Acoustic voice parameters such as fundamental frequency parameters,fre-quency variation parameters,amplitude variation parameters,formant parameters,and energy parameters were ex-tracted from the vowels/a/,/i/,and/u/.A one-way ANOVA test and Tukey's HSD post-hoc comparisons were conducted on the independent variables.Results There were no significant differences between smartphones and professional recording equipment in terms of fundamental frequency parameters such as median F0,mean F0,max F0 and min F0;frequency parameters such as jitter local,jitter local absolute,jitter rap,jitter ppq5,and jitter ddp;amplitude parameters such as shimmer local,shimmer local dB,shimmer apq3,shimmer apq5,shimmer apq11,and shimmer dda;and formant parameters such as F1,F2,F3,and F4.However,significant differences were found in energy parameters such as mean energy(F=31.171,P＜0.001),max energy(F=34.193,P＜0.001),and min energy(F=5.453,P＜0.001)between smartphones and professional recording equipment.Conclusion The smartphones using the"Active Health"screening app from the National Key Research and Development Program can replace professional recording equipment for voice research.However,caution should be exercised when selec-ting energy-related acoustic parameters.

Objective:To explore the prevalence and risk factors of insomnia in Chinese Air Force servicemen deployed to highland areas.Methods:A total of 718 Air Force servicemen deployed to Qinghai-Tibetan plateau were recruited at May 2024.Sleep quality was assessed with the Pittsburgh Sleep Quality Index.Social-demograph-ic,military service,and psychological characteristics were measured with a self-administered general question-naire.Bivariable and multivariable logistic regressions were performed to identify independent risk factors.Missing data were handled by the multiple imputation.Results:The average sleep duration was(6.9±1.2)h and the aver-age PSQI score was(5.9±4.1).Totally 53.8％of participants experienced clinically significant insomnia.The multivariable analysis revealed that age≥35(aOR=4.07,95％CI=1.11-17.76),stressful event(aOR=3.27,95％CI=2.00-5.49),dysfunctional sleep beliefs and attitudes(aOR=2.59,95％CI=1.75-3.85),and caffeine product usage(aOR=1.69,95％CI=1.17-2.43)were risk factors for insomnia,while Tibetan-indigenous ethnic(aOR=0.44,95％CI=0.20-0.91),higher perceived social support(aOR=0.96,95％CI=0.96-0.99),and positive coping style(aOR=0.96,95％CI=0.93-0.99)were protective factors.Conclusion:Air force service-men deployed to highland areas have sufficient sleep time,but reduced sleep quality.Age,exposed to stress event during deployment,dysfunctional sleep beliefs and attitudes,and caffeine product usage are risk factors for insomni-a,while Tibetan-indigenous ethnic,higher perceived social support and positive coping style act as protective fac-tors.

Objective To compare the consistency of voice parameters collected by commonly used smart-phone brands in China and professional recording equipment,and to study whether smartphones can be used for voice research.Methods A total of 67 normal subjects were selected for voice recording using six different smart-phone brands(via the"Active Health"screening APP from the National Key Research and Development Program)and professional recording equipment.Acoustic voice parameters such as fundamental frequency parameters,fre-quency variation parameters,amplitude variation parameters,formant parameters,and energy parameters were ex-tracted from the vowels/a/,/i/,and/u/.A one-way ANOVA test and Tukey's HSD post-hoc comparisons were conducted on the independent variables.Results There were no significant differences between smartphones and professional recording equipment in terms of fundamental frequency parameters such as median F0,mean F0,max F0 and min F0;frequency parameters such as jitter local,jitter local absolute,jitter rap,jitter ppq5,and jitter ddp;amplitude parameters such as shimmer local,shimmer local dB,shimmer apq3,shimmer apq5,shimmer apq11,and shimmer dda;and formant parameters such as F1,F2,F3,and F4.However,significant differences were found in energy parameters such as mean energy(F=31.171,P＜0.001),max energy(F=34.193,P＜0.001),and min energy(F=5.453,P＜0.001)between smartphones and professional recording equipment.Conclusion The smartphones using the"Active Health"screening app from the National Key Research and Development Program can replace professional recording equipment for voice research.However,caution should be exercised when selec-ting energy-related acoustic parameters.

Objective To observe the value of cervical lymph node ultrasound network model based on deep learning(DL)for diagnosing cervical lymph node metastasis of papillary thyroid carcinoma(PTC).Methods Totally 444 PTC patients with suspected cervical lymph node enlargement on ultrasonography were retrospectively enrolled and divided into metastasis group(n=253)and non-metastasis group(n=191)based on fine needle aspiration pathology and thyroglobulin detection.And 1 754 cervical lymph node ultrasonic images were divided into training set(n=1 404),validation set(n=175)and test set(n=175)at the ratio of 8∶1∶1.Ultrasonic features of cervical lymph nodes were extracted,then ultrasound network model was established using DL,and the diagnostic efficacy of this model for diagnosing lymph node metastasis was analyzed.Results The sensitivity,specificity,accuracy and area under the curve of ultrasound network model for diagnosing cervical lymph node metastasis of PTC was 93.06％,97.05％,95.04％and 0.858 in validation set,which was 86.14％,78.67％,88.49％and 0.828 in test set,respectively.Conclusion Cervical lymph node ultrasound network model based on DL was helpful for diagnosing cervical lymph node metastasis of PTC.

Objective:To investigate the risk factors associated with the development of venous thromboembolism with abdominal infection following pancreatic cancer surgery.Methods:A retrospective analysis was conducted on the clinical data of 70 patients with pancreatic cancer who underwent pancreaticoduodenectomy at the Department of Hepatobiliary Surgery of the First Affiliated Hospital of Air Force Medical University from March 2020 to August 2024, of which 31 were male and 39 were female, aged (58.5±6.8) years. Two groups were divided into two groups according to the occurrence of venous thromboembolism with abdominal infection in the postoperative period: the complication group ( n=28) and the non-complication group ( n=42). The general data, laboratory indexes and surgery-related indexes of the two groups were recorded, and logistic regression model was used to analyse the risk factors for the occurrence of venous thromboembolism with abdominal cavity infection after pancreaticoduodenectomy, and receiver operating characteristic (ROC) curve was used to assess the predictive value of the factors for the occurrence of venous thromboembolism with abdominal cavity infection after pancreatic cancer surgery. Results:Multifactorial analysis showed that the combination of diabetes mellitus ( OR=1.216, 95% CI: 1.225-1.657, P=0.035), occurrence of postoperative pancreatic fistula ( OR=1.292, 95% CI: 1.095-1.829, P=0.022), occurrence of postoperative biliary fistula ( OR=1.239, 95% CI: 1.171-2.618, P=0.005), high D-dimer ( OR=137.030, 95% CI: 3.214-5 843.128, P=0.010), prolonged operation time ( OR=1.058, 95% CI: 1.002-1.117, P=0.042), and low serum albumin ( OR=0.711, 95% CI: 0.508-0.995, P=0.046) in pancreatic cancer patients with a high risk of postoperative venous thromboembolism with the development of abdominal infection. ROC curve analysis showed that diabetes mellitus, pancreatic fistula, biliary fistula, D-dimer, duration of surgery, serum albumin, and co-prediction had certain predictive value for the occurrence of postoperative venous thromboembolism with abdominal infection after pancreatic cancer surgery, in which, the area under the curve for co-prediction in assessing the occurrence of postoperative venous thromboembolism with abdominal infection after pancreatic cancer surgery was 0.993 (95% CI: 0.982-1.000), with a sensitivity of 1.000 and a specificity of 0.929. Conclusion:Comorbid diabetes mellitus, pancreatic fistula, biliary fistula, D-dimer, prolonged duration of surgery, and serum albumin are independent risk factors for the development of postoperative venous thromboembolism with abdominal infection in patients with pancreatic cancer, all of which have diagnostic value and a high joint predictive value. The predictive value of co-prediction is relatively high.

OBJECTIVE To investigate the inhibitory effect and mechanism of the active components of Alpinia katsumadai （ACAK） on tumor xenograft growth and tumor angiogenesis of human pancreatic cancer PANC-1 cells in nude mice. METHODS A tumor xenograft model in nude mice was established using human pancreatic cancer PANC-1 cells. The mice were randomly divided into model control group （intragastric administration of 0.9% normal saline）， solvent control group （intragastric administration of 0.5% carboxymethyl cellulose sodium）， positive control group （intraperitoneal injection of 0.5% carboxymethyl cellulose sodium+bevacizumab suspension 5 mg/kg ）， and ACAK 50， 100， and 200 mg/kg groups （intragastric administration of 0.5% carboxymethyl cellulose sodium+ACAK suspension 50， 100， 200 mg/kg）. The administration was carried out for 5 consecutive days followed by a 2-day interval， and this cycle was repeated for a total duration of 28 days. The tumor volume （TV）， relative tumor volume （RTV）， and relative tumor proliferation rate （T/C） at various time points from day 1 to day 28 after drug administration were measured and calculated for each group of nude mice. After the drug administration， the tumor weights were measured， and microvessel density （MVD） in the tumor xenograft tissues of nude mice， as well as relative protein expression levels of vascular endothelial growth factor （VEGF） and its receptor [fas-like tyrosine kinase-1 （Flt-1）， kinase insert domain receptor （KDR）] were detected. RESULTS On the 24th day of ACAK administration，compared with the model control group， the TV and RTV （except for ACAK 50 and 100 mg/kg groups） of nude mice in the positive control group and ACAK dose groups were significantly decreased （P＜0.05 or P＜0.01）， and the T/C of ACAK dose groups showed a dose-dependent decrease； the microvascular distribution of nude mice in the positive control group and ACAK dose groups was relatively sparse， and the tumor weight （except for the ACAK 50 mg/kg group）， MVD， and relative expression levels of VEGF， KDR， and Flt-1 in the tumor xenograft tissues were significantly reduced （P＜0.05 or P＜0.01）. CONCLUSIONS ACAK has a good anti-pancreatic cancer effect， and its mechanism may be related to its inhibition of VEGF/ VEGFR signaling pathway， thereby inhibiting angiogenesis in pancreatic cancer.

Background Occupational stress has emerged as a critical public health concern affecting the physical and mental well-being of workers in the manufacturing sector. However, researchers typically evaluate its core driver—cumulative fatigue—using a crude binary “present/absent” variable, thereby overlooking the high-dimensional complexity and heterogeneity inherent in fatigue characteristics. This oversimplification constrains both the precision and predictive performance of occupational stress risk assessment model. Objective Leveraging a data-driven approach, to survey data on cumulative fatigue among manufacturing employees, and then use this new classification to develop and validate an occupational stress prediction model, with an ultimate aim of enhancing the accuracy and effectiveness of occupational stress assessment. Methods A set of cross-sectional survey data on 3871 manufacturing employees in 2021 were derived from the “Long working hours exposure and its adverse health effect risk assessment” program of the National Institute for Occupational Health and Poison Control, Chinese Center for Disease Control and Prevention. Occupational stress was assessed using the Core Occupational Stress Measurement Scale, while cumulative fatigue was evaluated via the Worker’s Self-Diagnostic Questionnaire for Fatigue Accumulation. Boruta method was applied to screen core variables from 20 occupational stress influencing factors. Dimensionality reduction of cumulative fatigue features was performed using a Bayesian sparse autoencoder (BASE), followed by comparison and application of clustering methods to achieve multi-class classification of the reduced features. Six machine learning classification models were then selected including logistic regression, support vector machine, decision tree, random forest, adaptive boosting, and lightweight gradient boosting machine (LightGBM) to construct and compare two occupational stress prediction models involving cumulative fatigue combined with ten other core variables: one utilizing the original binary cumulative fatigue label as a core variable, and another employing the novel fatigue classification proposed herein as a core variable. Results The positive rate of occupational stress among the manufacturing employees was 38.9%. The Boruta method identified 11 core variables of occupational stress: depressive symptoms, cumulative fatigue, age, length of service, tenure in current position, average weekly working hours, average daily overtime hours, average monthly income, low levels of exercise, life satisfaction, and sleep quality. The BSAE reduced the original cumulative fatigue factors to 12 dimensions, while the K-means clustering grouped the reduced fatigue features into three categories: no fatigue, moderate fatigue, and sever fatigue. Six occupational stress prediction models were constructed using the three-category labels of cumulative fatigue and ten additional factors. The results indicated that the LightGBM model performed best, achieving an area under the curve (AUC) of 0.78, an accuracy of 0.77, and an F1 score of 0.72. Compared with the model incorporating the traditional binary labels for cumulative fatigue, the best prediction AUC (0.72), accuracy (0.66), and F1 score (0.65) improved by 6%, 11%, and 7%, respectively. Conclusion Cumulative fatigue is a significant predictor of occupational stress among manufacturing employees. Applying data dimensionality reduction combined with three-class cluster analysis to characterize cumulative fatigue improves the performance of occupational stress prediction models. From a data-driven perspective, machine learning methods demonstrate strengths in processing complex datasets, capturing nonlinear relationships, and generating predictions based on key variables. This study therefore confirms that refined processing of core factors substantially enhances the predictive capability of machine learning models in assessing occupational stress risk in the manufacturing workforce.

ObjectiveTo assess the efficiency of a Sophora flavescens Ait (S. flavescens, Ku Shen)-soluble microneedle (SFA-MN) for improving skin lesion symptoms in mice with psoriasis.MethodsSFA-MNs were prepared using a two-mold molding process with 20% w/v polyvinylpyrrolidone and 15% w/v polyvinyl alcohol. The SFA-MNs were assessed for morphology, mechanical properties, in vitro dissolution, identification of components, and skin lesion improvement in imiquimod-induced psoriasis mice.ResultsThe SFA-MNs demonstrated good mechanical properties for efficiently penetrating the dermis, facilitating efficient drug delivery. Furthermore, they effectively inhibited mast cell levels in the dorsal lesion area of psoriasis mice and reduced the expression of the T-lymphocyte factor cluster of differentiation 3 and tumor necrosis factor-α. In addition, this system alleviated skin inflammation, splenic swelling, and thymic atrophy in the psoriasis-like mouse model. Seven major components were detected from SFA-MNs by comparison of the mass-to-nucleus ratios (m/z) of the secondary fragments N-methylcytisine, 5α, 9α-dihydroxymatrine, sophoramine, matrine, oxysophocarpine, oxymatrine, and kushenol O.ConclusionThe drug delivery strategy combining traditional herbal S. flavescens with soluble microneedle technology provides more targeted and effective immune regulation for treating psoriasis-like mice models, enabling enhanced therapeutic effects compared with the control group.