Objective: Machine learning (ML) has been reported to have better predictive capability than traditional statistical techniques. The aim of this study was to assess the efficacy of ML algorithms and logistic regression (LR) for predicting depressive symptoms during the COVID-19 pandemic. Methods: Analyses were carried out in a national cross-sectional study involving 21,916 participants. The ML algorithms in this study included random forest (RF), support vector machine (SVM), neural network (NN), and gradient boosting machine (GBM) methods. The performance indices were sensitivity, specificity, accuracy, precision, F1-score, and area under the receiver operating characteristic curve (AUC). Results: LR and NN had the best performance in terms of AUCs. The risk of overfitting was found to be negligible for most ML models except for RF, and GBM obtained the highest sensitivity, specificity, accuracy, precision, and F1-score. Therefore, LR, NN, and GBM models ranked among the best models. Conclusion Compared with ML models, LR model performed comparably to ML models in predicting depressive symptoms and identifying potential risk factors while also exhibiting a lower risk of overfitting.

Objective: To analyze the distribution characteristics of fluoride and trichloromethane in drinking water in rural schools in Guizhou Province and assess their health risks, so as to provide a scientific basis for ensuring the safety of drinking water in rural schools. Methods: During the dry season (March to May) and wet season (July to September) of 2020 to 2022, 788 rural primary and secondary schools in agricultural counties (districts) in Guizhou Province were selected for investigation by using a direct sampling method. A total of 1 566 drinking water samples were collected from these schools, and the mass concentrations of fluoride and trichloromethane in the water samples were detected. The Mann-Whitney U test was used for intergroup comparison, and a health risk assessment model was employed to evaluate the health risks of students oral intake of fluoride and trichloromethane. Results: From 2020 to 2022, the mass concentrations of fluoride and trichloromethane in the drinking water of rural schools in Guizhou Province all met the standards, and the ranges were no detection to 0.99 mg/L and (no detection to 0.06)×10 -3 mg/L, respectively. The mass concentrations of fluoride in dry and wet seasons were 0.05(0.05,0.10), 0.05(0.05,0.10) mg/L, the mass concentrations of trichloromethane were [0.02(0.02,1.00)]×10 -3 , [0.02(0.02,1.00)]×10 -3 mg/L, the mass concentrations of fluoride in factory water and terminal water were 0.05(0.05,0.05), 0.05(0.05,0.10) mg/L, and the differences were not statistically significant ( Z=-0.04, -0.88, - 0.98 , P >0.05). There was a statistically significant difference in the mass concentration of trichloromethane between factory water and peripheral water [0.02(0.02,0.02)×10 -3 , 0.02(0.02,1.05)×10 -3 mg/L]( Z=-2.16, P < 0.05 ). The non-carcinogenic risk assessment values for students oral exposure to fluoride and trichloromethane were in the range of 0.01(0.01,0.03)-0.03(0.03,0.06) and [0.26( 0.26 ,14.54)]×10 -4 -[0.52(0.52,48.62)]×10 -4 , respectively, all of which were at acceptable levels; the carcinogenic risk assessment values for oral exposure to trichloromethane were in the range of [0.08(0.08, 4.51 )]×10 -7 -[0.16(0.16,15.07)]×10 -7 , indicating a low risk. Conclusions The health risks of students expore to fluoride and trichloromethane in drinking water in rural schools of Guizhou Province are low. It is necessary to strengthen the standardized management of disinfection in some rural drinking water projects and the monitoring of fluoride in water sources to reduce the exposure risk to children.

OBJECTIVE To establish management index system for rational drug use of key monitoring drugs， and provide reference for the management of key monitoring drugs in the hospitals. METHODS First， the management index system for rational drug use of key monitoring drugs was drafted by collecting the evidence from related medical literature. Next， using a modified Delphi method， twenty experienced experts from the fields of pharmacy， medical practice， healthcare insurance， and finance were selected to participate in two rounds of questionnaire consultations. Based on the expert enthusiasm coefficient， authority coefficient， degree of opinion concentration， and degree of coordination， the final indicators were determined to establish a management index system for rational drug use of key monitored drugs in medical institutions. RESULTS The expert enthusiasm coefficients reached 100% in both rounds of consultation. In first-level， second-level and third-level indicators， the authority coefficients of experts were 0.89， 0.86 and 0.87， and coordination coefficients of the experts in importance score were 0.300 （P＜ 0.05）， 0.125 （P＜0.05） and 0.139 （P＜0.05）， respectively. The average score for the importance of all indicators reached over 3.5， in which the full score ratio ranged from 35% to 100%. Except that the variation coefficient of a third-level indicator “number of specifications purchased for key monitored drugs” was 0.26， the variation coefficients of rest indicators were less than or equal to 0.25. Based on the results of expert consultation， final version of the management index system established in this study， including two first-level indicators （drug procurement and use， and rational drug use）， five second-level indicators （such as the accessibility， cost-effectiveness） and twenty third-level indicators （such as the number of specifications purchased for key monitored drugs， the increase in the cost of key monitored drugs）. CONCLUSIONS The management index system established in this study possesses high reliability and strong operability， and may provide a reference for the management of key monitoring drugs in the hospitals.

Objective: Machine learning (ML) has been reported to have better predictive capability than traditional statistical techniques. The aim of this study was to assess the efficacy of ML algorithms and logistic regression (LR) for predicting depressive symptoms during the COVID-19 pandemic. Methods: Analyses were carried out in a national cross-sectional study involving 21,916 participants. The ML algorithms in this study included random forest (RF), support vector machine (SVM), neural network (NN), and gradient boosting machine (GBM) methods. The performance indices were sensitivity, specificity, accuracy, precision, F1-score, and area under the receiver operating characteristic curve (AUC). Results: LR and NN had the best performance in terms of AUCs. The risk of overfitting was found to be negligible for most ML models except for RF, and GBM obtained the highest sensitivity, specificity, accuracy, precision, and F1-score. Therefore, LR, NN, and GBM models ranked among the best models. Conclusion Compared with ML models, LR model performed comparably to ML models in predicting depressive symptoms and identifying potential risk factors while also exhibiting a lower risk of overfitting.

Objective: Machine learning (ML) has been reported to have better predictive capability than traditional statistical techniques. The aim of this study was to assess the efficacy of ML algorithms and logistic regression (LR) for predicting depressive symptoms during the COVID-19 pandemic. Methods: Analyses were carried out in a national cross-sectional study involving 21,916 participants. The ML algorithms in this study included random forest (RF), support vector machine (SVM), neural network (NN), and gradient boosting machine (GBM) methods. The performance indices were sensitivity, specificity, accuracy, precision, F1-score, and area under the receiver operating characteristic curve (AUC). Results: LR and NN had the best performance in terms of AUCs. The risk of overfitting was found to be negligible for most ML models except for RF, and GBM obtained the highest sensitivity, specificity, accuracy, precision, and F1-score. Therefore, LR, NN, and GBM models ranked among the best models. Conclusion Compared with ML models, LR model performed comparably to ML models in predicting depressive symptoms and identifying potential risk factors while also exhibiting a lower risk of overfitting.

Objective: Machine learning (ML) has been reported to have better predictive capability than traditional statistical techniques. The aim of this study was to assess the efficacy of ML algorithms and logistic regression (LR) for predicting depressive symptoms during the COVID-19 pandemic. Methods: Analyses were carried out in a national cross-sectional study involving 21,916 participants. The ML algorithms in this study included random forest (RF), support vector machine (SVM), neural network (NN), and gradient boosting machine (GBM) methods. The performance indices were sensitivity, specificity, accuracy, precision, F1-score, and area under the receiver operating characteristic curve (AUC). Results: LR and NN had the best performance in terms of AUCs. The risk of overfitting was found to be negligible for most ML models except for RF, and GBM obtained the highest sensitivity, specificity, accuracy, precision, and F1-score. Therefore, LR, NN, and GBM models ranked among the best models. Conclusion Compared with ML models, LR model performed comparably to ML models in predicting depressive symptoms and identifying potential risk factors while also exhibiting a lower risk of overfitting.

Objective: Machine learning (ML) has been reported to have better predictive capability than traditional statistical techniques. The aim of this study was to assess the efficacy of ML algorithms and logistic regression (LR) for predicting depressive symptoms during the COVID-19 pandemic. Methods: Analyses were carried out in a national cross-sectional study involving 21,916 participants. The ML algorithms in this study included random forest (RF), support vector machine (SVM), neural network (NN), and gradient boosting machine (GBM) methods. The performance indices were sensitivity, specificity, accuracy, precision, F1-score, and area under the receiver operating characteristic curve (AUC). Results: LR and NN had the best performance in terms of AUCs. The risk of overfitting was found to be negligible for most ML models except for RF, and GBM obtained the highest sensitivity, specificity, accuracy, precision, and F1-score. Therefore, LR, NN, and GBM models ranked among the best models. Conclusion Compared with ML models, LR model performed comparably to ML models in predicting depressive symptoms and identifying potential risk factors while also exhibiting a lower risk of overfitting.

Aim To investigate the effect of Xuefu Zhuyu decoction on transforming growth factor-β1(TGF-β1 ) -induced endothelial-to-mesenchymal transition (EndMT) of pulmonary microvascular endothelial cells ( PMVEC), and further analyze the mechanism related to the TGF-β1/Smad signaling pathway. Method To construct an EndMT cell model, PMVEC was treated with TGF-β1 (5 μg · L

italic>Anoectochilus roxburghii liquid (spray, a hospital preparation of Wu Mengchao Hepatobiliary Hospital of Fujian Medical University) has shown a good clinical treatment effect during the COVID-19 pandemic, but its material basis and mechanism of action are still unclear. In this study, network pharmacology and molecular docking methods were used to predict the molecular mechanism of A. roxburghii liquid against COVID-19, and pharmacodynamic experiments in vitro were conducted to study the interaction between the current targets with clear preventive and therapeutic effects and the key components of A. roxburghii liquid. UPLC-MS and database were used to compare and analyze the active ingredients in the liquid, and 17 potential active ingredients with good drug-like properties were screened by in vivo pharmacokinetics process in SwissADME database. SwissTargetPrediction and GeneCards were searched to find 93 common targets. Cytoscape 3.8.2 software was used to construct the "component-target" network map, and the Metascape platform was used for gene function annotation and pathway enrichment analysis. It was found that the extract could regulate the positive response to external stimuli, inflammatory response, cytokine production and other biological processes by binding the active ingredients such as isorhamnetin, kaempferol, luteolin, quercetin and apigenin to the common targets (NOS3, MPO, MMP3, etc.), and play an anti-COVID-19 role. In the angiotensin-converting enzyme 2 (ACE2) activity inhibition assay, it was found that the stock solution of A. roxburghii liquid (for spray), and the supernatant after removing polysaccharides (mainly containing flavonoids) could to some extent inhibit the activity of ACE2. Crucially, in the experiment of 2019-nCOV-S pseudovirus infecting HEK-293T-ACE2 cells, we found that A. roxburghii liquid may exert anti-COVID-19 effects by blocking the binding of SARS-CoV-S protein to ACE2.

ObjectiveTo improve the response capabilities to disasters and prevent major epidemics， it is of practical use to study the capability evaluation system of the national emergency medical rescue team that combines theoretical training and practical exercises， to enhance the overall quality of the teams. MethodsFirst， a capability assessment system for the national emergency medical rescue team was constructed based on the INSARAG External Classification （IEC） standards of the national emergency medical rescue team. Then， based on the outcome based education （OBE） concept， we conducted innovative research on the curriculum design and exercise programs for team building and member training. Finally， an empirical analysis was conducted on the effectiveness of the evaluation system and training exercises based on the statistical analysis of the comprehensive quality evaluation of the Shanghai national emergency medical rescue team from 2020 to 2023， as well as the empirical analysis of the rescue exercise on the Cruise of spectrum. ResultsBased on the linear regression analysis of each core competency indicators， the five core competencies in the evaluation system， including rescue skills， medical and health knowledge， disaster coping ability， team cooperation ability， and mental resilience training， were positively correlated with the cumulative number of trainings （r=0.71， r=0.76， r=0.81， r=0.84， r=0.96，all P<0.05）， indicating that the training was effective and the course design was reasonable. Empirical cases showed that the three-dimensional rescue drill model had remarkable results in the actual combat application and ability improvement of team members. ConclusionThe training courses and drills designed based on the three-level assessment system are effective in improving the comprehensive capabilities of the national emergency medical rescue team.