Objective To investigate plasma protein expression changes in heatstroke patients using proteomics technology and to identify reliable prognostic biomarkers.Methods A retrospective analysis was conducted on 20 heatstroke patients hospitalized at the 908th Hospital of the Chinese PLA Joint Logistics Support Force from July 2022 to February 2024.Patients were divided into survival(n=16)and death groups(n=4)based on 28-day outcomes.Fasting venous blood samples were collected from both groups for proteomic analysis.Liquid chromatography-mass spectrometry(LC-MS/MS)was used to identify and screen differentially expressed proteins.Kyoto Encyclopedia of Genes and Genomes(KEGG)and Gene Ontology(GO)enrichment analyses were performed.LASSO regression was applied to screen key prognostic biomarkers,and receiver operating characteristic(ROC)curve analysis was used to evaluate their predictive value.Results A total of 27 upregulated and 90 downregulated proteins were identified between the two groups.These proteins were primarily involved in biological processes such as immune responses,complement activation,and metabolic processes.LASSO regression analysis indicated that actin-related protein 2(ARP2),cysteine-rich scavenger receptor type 1 M130(CD163),and catalase(CAT)could serve as effective biomarkers for evaluating heatstroke prognosis.ROC curve analysis demonstrated that ARP2 had higher diagnostic efficacy(AUC=0.98,sensitivity=0.80,specificity=1.00)compared to CD163(AUC=0.94,sensitivity=0.76,specificity=1.00),CAT(AUC=0.96,sensitivity=0.67,specificity=1.00),and acute physiology and chronic health evaluation Ⅱ(APACHE Ⅱ)score(AUC=0.79,sensitivity=0.44,specificity=1.00).Conclusion Elevated plasma ARP2 level has good clinical value for predicting poor prognosis in heatstroke patients.

Objective To explore the pathogenesis of Alzheimer's disease by examining the effects of dihydroartemisinin(DHA)on cognitive behavior,hippocampal,cerebral cortex and retinal cell morphology,β-amyloid(Aβ)and autophagy-related proteins in 5×FAD mice.Methods Twenty 5×FAD mice and 5 wild type(WT)mice were selected,all of which were female.The 5×FAD mice were randomly divided into model(M)group,donepezil(D)group,low-dose DHA(DHA-L)group,and high-dose DHA(DHA-H)group.The WT and M groups were not treated,and the D group was given donepezil 0.1 mg/kg per day.DHA-L group and DHA-H group were given 10 mg/kg and 20 mg/kg DHA per day,respectively.Group D,group DHA-L and group DHA-H were given intragastric administration once a day for 3 months.The changes of in cognitive behavior were measured by Morris experiment.HE staining was used to observe the arrangement and morphology of nerve cells in cerebral cortex,hippocampus and retina.The expressions of Aβ protein in cerebral cortex,hippocampus and retina were detected by immunohistochemistry.Western blotting detected the expression of autophagy related proteins(LC3-Ⅰ,LC3-Ⅱ,Beclin-1,P62,β-actin).Results The DHA-H group and the D group exhibited more frequent adoption of both linear and trending exploration routes.Compared to the model group,significant differences in the contents of Aβ in the hippocampal CA1,cerebral cortex S1,and retinal were observed(P＜0.0001)in the other four groups.The analysis also showed significant differences in autophagy-associated proteins between the DHA-L,DHA-H,and model groups(P＜0.01).Conclusion DHA improves cognitive function and increases the number of nerve cells in mice.It also reduces Aβ content in the cerebral cortex,hippocampus,and retina,along with improving autophagy-associated protein deposition in mice.

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.

Microplastics are widely present in various environments such as water bodies,land,and atmosphere,which pose threats to the ecological environment and human health through transmission and accumulation in the food chain.The existing detection techniques for microplastics face challenges such as complex preparation procedure of samples,low efficiency in processing large batches of samples,and difficulties in handling complex samples.Therefore,there is an urgent need for rapid and efficient detection techniques suitable for complex microplastics samples in the field of environmental monitoring.Raman spectroscopy,known for its advantages such as rapidity,accuracy,high sensitivity,non-destructiveness,and non-contact,demonstrates great application potential in detection of microplastics.Deep learning,an artificial intelligence method known for its large-scale data processing,nonlinear modeling and automatic feature extraction capabilities,is receiving increasing attention in the analysis of Raman spectroscopy signals.The application of deep learning-based Raman spectroscopy has significantly improved performance indicators such as detection efficiency and accuracy.This article introduced the existing Raman enhancement techniques,summarized the deep learning methods applied in Raman spectroscopy signal analysis,reviewed the recent research and application progress of deep learning-based Raman spectroscopy in detection of microplastics,and finally discussed the challenges and future prospects of deep learning-based Raman spectroscopy in detection of microplastics.

The single photon counting imaging detector based on microchannel plate(MCP)has the characteristics of high sensitivity and low dark count rate,and has been applied to the optical remote sensing detection of weak ultraviolet spectral signals in space.In this work,by using planar array single photon counting imaging detector as the detector,flat-field concave grating as the splitter,and electrodeless discharge lamp(EDL)as the excitation light source,a dispersion detection system suitable for hydride generation-atomic fluorescence spectrometry(HG-AFS)was developed.The wavelength calibration of the system was carried out,and the negative high pressure and EDL stability time of the planar array single photon counting imaging detector were analyzed and optimized.The characteristic emission spectral lines of As and Bi elements excited in the wavelength range of 180-320 nm were analyzed,and the scattering interference in the wavelength range of 257.3-306.7 nm was discussed.The results showed that the AFS dispersion detection system based on the planar array single photon counting imaging detector could detect and analyze the HG-AFS fluorescence signal initially,and the influence of scattering interference on the detection results was effectively avoided.The system had the advantages including simple structure,no refrigeration and temperature control,no moving parts and simultaneous measurement of multi-band.

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.

Objective:To investigate the prevalence of sexual drug use in men who have sex with men (MSM) in Chengdu and analyze the relationship between depression and sexual drug use.Methods:A total of 1 277 MSM were recruited between November 2021 and May 2022. Questionnaire was used to collect information about their demographic characteristics, depression status and sexual drug use behavior. Univariate and multivariate logistic regression models were used to analyze the relationship between depression and sexual drug use.Results:In the 1 277 MSM, assessment identified 503 mild depression cases (39.4%), 196 moderate depression cases (15.3%) and 171 severe depression cases (13.4%) and 444 MSM (34.8%) reported sexual drug use in the past 6 months. Multivariate logistic regression models showed that compared with non-depression, mild depression (a OR=1.67,95% CI:1.24-2.23), moderate depression (a OR=1.50,95% CI: 1.02-2.19) and severe depression (a OR=1.56,95% CI:1.04-2.32) were positively associated with sexual drug use. Conclusions:The prevalence of depression and sexual drug use were high in MSM in Chengdu. There was a positive correlation between depression and sexual drug use. It is necessary to pay close attention to depression and sexual drug use and conduct targeted intervention in MSM.