Objective: To investigate the basic situation of pre-analytical quality management in blood station laboratories in North China, and to provide baseline data for promoting the homogenization and standardization of these pre-analytical processes in each blood station laboratory. Methods: A cross-sectional status survey was designed based on the quality management regulations of blood stations, ISO15189 standards and relevant quality management requirements. This survey covering various aspects including laboratory general situation, sample collection and temporary storage, transportation, reception, and quality continuous improvement situations. Data analysis was performed on the survey results of each laboratory. Results: All the 38 blood station laboratories in North China had established a pre-analytical quality management system framework and implemented basic pre-analytical quality control activities; however, there were differences in implementation. 1) Among the 12 basic quality items, 3 items were monitored by all the investigated laboratories (100%), 6 items were monitored by the vast majority of laboratories (about 90%), and 3 items were monitored by a portion of laboratories (about 60%). There were no significant differences in the monitoring index among the three regions and among different types of laboratories (P>0.05). 2) Among the total of 26 items in the three key processes before testing (sample collection and storage, transportation, reception and processing), 12 items were monitored by all laboratories (100%), 11 items were monitored by the vast majority of laboratories (about 90%), and 3 items were monitored by a portion of laboratories (about 75%). There were no significant differences in monitoring index among different regions and types of laboratories (P>0.05). Conclusion: This survey provides a reference and basis for the gap analysis of the pre-analytical process quality management in 38 blood station laboratories across North China. It facilitates laboratories in identifying pre-analytical quality problems, resolving problems, preventing errors, and ensuring that the quality of blood samples before testing meets the established requirements. It lays a foundation for the homogenization of pre-analytical quality management in regional blood stations.

Objective: To determine the frequency and main reasons of unqualified samples by analyzing the quality of pre-analytical samples in blood stations in North China, thereby providing a reference and basis for gap analysis in the implementation of pre-analytical process quality management for participating laboratories and ensuring that only high-standard and high-quality blood samples proceed to testing. Methods: Data on the quality of pre-analytical samples from blood station laboratories in North China was collected via questionnaire. Statistical analysis were performed on: 1) the basic information of samples quality monitoring in the laboratories; 2) the distribution of the overall pre-analytical unqualified rate of samples and the pre-analytical unqualified rate of samples in each laboratory; 3) the distribution of reasons for sample disqualification. Results: 1) The overall pre-analytical unqualified rate of samples in blood station laboratories in North China was 4.55, with a total sigma level of 5.39σ. The 25th, 50th and 75th percentiles (P25, P50, P75) for the total unqualified rate were 0.00, 1.10 and 5.96, respectively. The corresponding percentiles for the Sigma level were 5.34σ, 5.71σ, and 6.00σ, respectively. The pre-analytical unqualified rate of serological and nucleic acid samples (4.89 vs 4.22) showed a significant difference (χ =9.575, P<0.05). 2) The average unqualified rate of samples in region A, B and C was 1.71, 9.50 and 12.64 (χ =1 590.721, P<0.05), and the sigma level was 5.66σ, 5.21σ and 5.16σ, respectively. 3) The main reasons for unqualified serological samples were chylous blood (72.65%), hemolysis (17.39%), abnormal hematocrit (5.80%), and insufficient volume (3.50%). The main reasons for the unqualified nucleic acid samples were chylous blood (78.26%), hemolysis (8.84%), failure to centrifuge as required (5.01%), abnormal hematocrit (4.66%), and insufficient volume (1.92%). Conclusion: In North China, the quality indicators for the pre-analytical processes in blood station laboratories are generally well-managed. Laboratories in region A outperformed the national average in pre-analytical specimen quality control. However, participating laboratories exhibit gaps in implementing pre-analytical quality management. Through effective analysis of pre-analytical process quality metrics and inter-laboratory comparisons, laboratories can identify discrepancies and address shortcomings. By establishing clear quality objectives, they can achieve continuous improvement and ensure the validity of test results.

Objective: To investigate the assessment criteria and subsequent handling practices of hemolytic and lipemic blood samples before testing in blood screening laboratories in North China, and to provide data to support the standardization of their management in blood station laboratories. Methods: Data on the preanalytical management of hemolytic and lipemic samples from 38 laboratories were collected. The details of management on the criteria and verificatioon for assessment, the assessment methods, and subsequent handling procedures of hemolytic and lipemic samples in blood station laboratories were analyzed. Results: 1) All 38 blood station laboratories monitored serological and nucleic acid samples for hemolysis and lipemia in pre-analytical phase. 2) The criteria and methods for assessing hemolytic and lipemic samples varied among the laboratories of the 38 blood stations. 15 laboratories (39.47%) followed manufacturer's instructions, 9 laboratories (23.68%) formulated their own criteria, and 14 laboratories (36.84%) referred to the criteria of other laboratories. 16 laboratories (42.11%) verified the criteria for assessing hemolytic and lipemic samples, with significant variations in verification rate across laboratories from different regions (P<0.05). For the assessment methods, visual inspection was used by 28 laboratories (73.68%) for hemolytic samples and by 27 laboratories (71.05%) for lipemic samples; the colorimetric card method was used by 10 laboratories (26.32%) for assessing both hemolytic and lipemic samples; the instrumental method was used by 1 laboratory (2.63%) for assessing lipemic samples.3) The handling procedures for hemolytic and lipemic samples varied significantly and followed a gradient distribution pattern among 38 laboratories (including accepting samples for testing, accepting samples for concession testing, re-collecting samples, and rejecting samples and halting testing). With increasing severity of hemolysis and lipemia, more laboratories halted testing, and relatively fewer laboratories accepted samples for normal testing. 5 laboratories (13.16%) applied different handling procedures on serological and nucleic acid samples. Conclusion: This survey provides a reference and basis for analyzing gaps in the management of hemolytic and lipemic samples during the preanalyical phase in blood station laboratories in North China. It enables laboratories to identify the problems and deficiencies in the management of hemolytic and lipemic samples, to ensure preanalytical samples quality meets the established requirements, and to lay a foundation for promoting the homogenization and standardization of the regional sample quality management mode.

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 distribution and antimicrobial resistance profiles of clinically isolated Haemophilus influenzae and Moraxella catarrhalis in hospitals across China from 2015 to 2021,and provide evidence for rational use of antimicrobial agents.Methods Data of H.influenzae and M.catarrhalis strains isolated from 2015 to 2021 in CHINET program were collected for analysis,and antimicrobial susceptibility testing was performed by disc diffusion method or automated systems according to the uniform protocol of CHINET.The results were interpreted according to the CLSI breakpoints in 2022.Beta-lactamases was detected by using nitrocefin disk.Results From 2015 to 2021,a total of 43 642 strains of Haemophilus species were isolated,accounting for 2.91％of the total clinical isolates and 4.07％of Gram-negative bacteria in CHINET program.Among the 40 437 strains of H.influenzae,66.89％were isolated from children and 33.11％were isolated from adults.More than 90％of the H.influenzae strains were isolated from respiratory tract specimens.The prevalence of β-lactamase was 53.79％in H.influenzae strains.The H.influenzae strains isolated from children showed higher resistance rate than the strains isolated from adults.Overall,779 strains of H.influenzae did not produce β-lactamase but were resistant to ampicillin(BLNAR).Beta-lactamase-producing strains showed significantly higher resistance rates to these antimicrobial agents than the β-lactamase-nonproducing strains.Of the 16 191 M.catarrhalis strains,80.06％were isolated from children and 19.94％isolated from adults.M.catarrhalis strains were mostly susceptible to both amoxicillin-clavulanic acid and cefuroxime,evidenced by resistance rate lower than 2.0％.Conclusions The emergence of antibiotic-resistant H.influenzae due to β-lactamase production poses a challenge for clinical anti-infective treatment.Therefore,it is very important to implement antibiotic resistance surveillance for H.influenzae and guide rational antibiotic use.All local clinical microbiology laboratories should actively improve antibiotic susceptibility testing and strengthen antibiotic resistance surveillance for H.influenzae.

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.

BACKGROUND: Hemorrhagic stroke (HS) is associated with significant disability and mortality. However, the relationship between meteorological factors and hemorrhagic stroke, as well as the potential moderating role of these factors, remains unclear. METHODS: Daily data on HS, air pollution, and meteorological conditions were collected from January 2015 to December 2021 in Ganzhou to analyze the relationship between meteorological factors and HS admissions. This analysis employed a time-stratified case-crossover design in conjunction with a distributional lag nonlinear model. Additionally, a bivariate response surface modelling was utilized to further investigate the interaction between meteorological factors and particulate matter. The study also stratified the analyses by gender and age. To investigate the potential impact of extreme weather conditions on HS, this study defined the 97.5th percentile as representing extremely high weather conditions, while the 2.5th percentile was classified as extremely low. RESULTS: In single-day lags, the risk of admissions for HS was significantly associated with extremely low temperature (lag 1-2 and lag 13-14), extremely low humidity (lag 1 and lag 9-12), and extremely high precipitation (lag 2-7). Females exhibited greater susceptibility to extremely low temperature than males within the single-day lag pattern in the subcomponent layer, with a maximum relative risk (RR) that was 7% higher. In the cumulative lag analysis, the risk of HS admissions was significantly associated with extremely high temperature (lag 0-8∼lag 0-14), extremely low humidity (lag 0-2∼lag 0-14), and extremely high precipitation (lag 0-4∼lag 0-14). Within the cumulative lag day structure of the subcomponent layer, both extremely low and extremely high temperature had a more pronounced effect on females and aged ≥65 years. The risk of HS admissions was positively associated with extremely high barometric pressure in the female subgroups (lag 0-1 and lag 0-2). The highest number of HS admissions occurred when high PM_2.5 concentrations coexisted with low precipitation. CONCLUSIONS Meteorological factors were significantly associated with the risk of hospital admissions for HS. Individuals who were female and aged ≥65 years were found to be more susceptible to these meteorological influences. Additionally, an interaction was observed between airborne particulate matter and meteorological factors. These findings contributed new evidence to the association between meteorological factors and HS.
China/epidemiology* ; Humans ; Female ; Male ; Aged ; Middle Aged ; Cross-Over Studies ; Hospitalization/statistics & numerical data* ; Adult ; Hemorrhagic Stroke/etiology* ; Meteorological Concepts ; Weather ; Particulate Matter/analysis* ; Air Pollution/adverse effects* ; Environmental Exposure/adverse effects* ; Aged, 80 and over ; Young Adult

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.

The working principle of Bausch&Lomb BL11110 phacoemulsification system was described.Three cases of typical faults of the phacoemulsification system were introduced,and the causes were analyzed,then the maintenance measures were given accordingly.References were provided for diagnosing and eliminating the faults of the phacoemulsification system.[Chinese Medical Equipment Journal,2025,46(4):118-120]