1.Comparison of Logistic Regression and Machine Learning Approaches in Predicting Depressive Symptoms: A National-Based Study
Xing-Xuan DONG ; Jian-Hua LIU ; Tian-Yang ZHANG ; Chen-Wei PAN ; Chun-Hua ZHAO ; Yi-Bo WU ; Dan-Dan CHEN
Psychiatry Investigation 2025;22(3):267-278
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.
2.Comparison of Logistic Regression and Machine Learning Approaches in Predicting Depressive Symptoms: A National-Based Study
Xing-Xuan DONG ; Jian-Hua LIU ; Tian-Yang ZHANG ; Chen-Wei PAN ; Chun-Hua ZHAO ; Yi-Bo WU ; Dan-Dan CHEN
Psychiatry Investigation 2025;22(3):267-278
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.
3.Comparison of Logistic Regression and Machine Learning Approaches in Predicting Depressive Symptoms: A National-Based Study
Xing-Xuan DONG ; Jian-Hua LIU ; Tian-Yang ZHANG ; Chen-Wei PAN ; Chun-Hua ZHAO ; Yi-Bo WU ; Dan-Dan CHEN
Psychiatry Investigation 2025;22(3):267-278
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.
4.Comparison of Logistic Regression and Machine Learning Approaches in Predicting Depressive Symptoms: A National-Based Study
Xing-Xuan DONG ; Jian-Hua LIU ; Tian-Yang ZHANG ; Chen-Wei PAN ; Chun-Hua ZHAO ; Yi-Bo WU ; Dan-Dan CHEN
Psychiatry Investigation 2025;22(3):267-278
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.
5.Comparison of Logistic Regression and Machine Learning Approaches in Predicting Depressive Symptoms: A National-Based Study
Xing-Xuan DONG ; Jian-Hua LIU ; Tian-Yang ZHANG ; Chen-Wei PAN ; Chun-Hua ZHAO ; Yi-Bo WU ; Dan-Dan CHEN
Psychiatry Investigation 2025;22(3):267-278
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.
6.Mechanism related to bile acids metabolism of liver injury induced by long-term administration of emodin.
Jing-Zhuo TIAN ; Lian-Mei WANG ; Yan YI ; Zhong XIAN ; Nuo DENG ; Yong ZHAO ; Chun-Ying LI ; Yu-Shi ZHANG ; Su-Yan LIU ; Jia-Yin HAN ; Chen PAN ; Chen-Yue LIU ; Jing MENG ; Ai-Hua LIANG
China Journal of Chinese Materia Medica 2025;50(11):3079-3087
Emodin is a hydroxyanthraquinone compound that is widely distributed and has multiple pharmacological activities, including anti-diarrheal, anti-inflammatory, and liver-protective effects. Research indicates that emodin may be one of the main components responsible for inducing hepatotoxicity. However, studies on the mechanisms of liver injury are relatively limited, particularly those related to bile acids(BAs) metabolism. This study aims to systematically investigate the effects of different dosages of emodin on BAs metabolism, providing a basis for the safe clinical use of traditional Chinese medicine(TCM)containing emodin. First, this study evaluated the safety of repeated administration of different dosages of emodin over a 5-week period, with a particular focus on its impact on the liver. Next, the composition and content of BAs in serum and liver were analyzed. Subsequently, qRT-PCR was used to detect the mRNA expression of nuclear receptors and transporters related to BAs metabolism. The results showed that 1 g·kg~(-1) emodin induced hepatic damage, with bile duct hyperplasia as the primary pathological manifestation. It significantly increased the levels of various BAs in the serum and primary BAs(including taurine-conjugated and free BAs) in the liver. Additionally, it downregulated the mRNA expression of farnesoid X receptor(FXR), retinoid X receptor(RXR), and sodium taurocholate cotransporting polypeptide(NTCP), and upregulated the mRNA expression of cholesterol 7α-hydroxylase(CYP7A1) in the liver. Although 0.01 g·kg~(-1) and 0.03 g·kg~(-1) emodin did not induce obvious liver injury, they significantly increased the level of taurine-conjugated BAs in the liver, suggesting a potential interference with BAs homeostasis. In conclusion, 1 g·kg~(-1) emodin may promote the production of primary BAs in the liver by affecting the FXR-RXR-CYP7A1 pathway, inhibit NTCP expression, and reduce BA reabsorption in the liver, resulting in BA accumulation in the peripheral blood. This disruption of BA homeostasis leads to liver injury. Even doses of emodin close to the clinical dose can also have a certain effect on the homeostasis of BAs. Therefore, when using traditional Chinese medicine or formulas containing emodin in clinical practice, it is necessary to regularly monitor liver function indicators and closely monitor the risk of drug-induced liver injury.
Emodin/administration & dosage*
;
Bile Acids and Salts/metabolism*
;
Animals
;
Male
;
Liver/injuries*
;
Chemical and Drug Induced Liver Injury/genetics*
;
Drugs, Chinese Herbal/adverse effects*
;
Humans
;
Rats, Sprague-Dawley
;
Mice
;
Rats
7.Air pollution exposure associated with decline rates in skeletal muscle mass and grip strength and increase rate in body fat in elderly: a 5-year follow-up study.
Chi-Hsien CHEN ; Li-Ying HUANG ; Kang-Yun LEE ; Chih-Da WU ; Shih-Chun PAN ; Yue Leon GUO
Environmental Health and Preventive Medicine 2025;30():56-56
BACKGROUND:
The effect of air pollution on annual change rates in grip strength and body composition in the elderly is unknown.
OBJECTIVES:
This study evaluated the effects of long-term exposure to ambient air pollution on change rates of grip strength and body composition in the elderly.
METHODS:
In the period 2016-2020, grip strength and body composition were assessed and measured 1-2 times per year in 395 elderly participants living in the Taipei basin. Exposure to ambient fine particulate matters (PM2.5), nitric dioxide (NO2), and ozone (O3) from 2015 to 2019 was estimated using a hybrid Kriging/Land-use regression model. In addition, long-term exposure to carbon monoxide (CO) was estimated using an ordinary Kriging approach. Associations between air pollution exposures and annual changes in health outcomes were analyzed using linear mixed-effects models.
RESULTS:
An inter-quartile range (4.1 µg/m3) increase in long-term exposure to PM2.5 was associated with a faster decline rate in grip strength (-0.16 kg per year) and skeletal muscle mass (-0.14 kg per year), but an increase in body fat mass (0.21 kg per year). The effect of PM2.5 remained robust after adjustment for NO2, O3 and CO exposure. In subgroup analysis, the PM2.5-related decline rate in grip strength was greater in participants with older age (>70 years) or higher protein intake, whereas in skeletal muscle mass, the decline rate was more pronounced in participants having a lower frequency of moderate or strenuous exercise. The PM2.5-related increase rate in body fat mass was higher in participants having a lower frequency of strenuous exercise or soybean intake.
CONCLUSIONS
Among the elderly, long-term exposure to ambient PM2.5 is associated with a faster decline in grip strength and skeletal muscle mass, and an increase in body fat mass. Susceptibility to PM2.5 may be influenced by age, physical activity, and dietary protein intake; however, these modifying effects vary across different health outcomes, and further research is needed to clarify their mechanisms and consistency.
Humans
;
Hand Strength
;
Aged
;
Male
;
Female
;
Environmental Exposure/adverse effects*
;
Follow-Up Studies
;
Taiwan
;
Air Pollution/adverse effects*
;
Particulate Matter/adverse effects*
;
Muscle, Skeletal/drug effects*
;
Air Pollutants/adverse effects*
;
Ozone/adverse effects*
;
Aged, 80 and over
;
Adipose Tissue/drug effects*
;
Body Composition/drug effects*
;
Nitrogen Dioxide/adverse effects*
8.Additional role of low-density lipoprotein cholesterol on the risk of osteoporosis in men with or without coronary heart disease: a real-world longitudinal study.
Jing ZENG ; Zi-Mo PAN ; Ting LI ; Ze-Yu CHEN ; Xiao-Yan CAI ; Mei-Liang GONG ; Xin-Li DENG ; Sheng-Shu WANG ; Nan LI ; Miao LIU ; Chun-Lin LI
Journal of Geriatric Cardiology 2025;22(2):219-228
BACKGROUND:
Early control of low-density lipoprotein cholesterol (LDL-C) is crucial for reducing the progress of cardiovascular disease. However, its additional role to the risk of primary osteoporosis in men with coronary heart disease was inconclusive. Our study aims to determine the association of LDL-C and its trajectories for osteoporosis risk in the middle-aged and aged men of China.
METHODS:
The retrospective cohort study of 1546 men aged 69.74 ± 11.30 years conducted in Beijing, China from 2015 to 2022. And the incidence of primary osteoporosis was annually recorded. LDL-C trajectories were further identified by latent class growth model using repeated measurements of LDL-C. The association of baseline LDL-C for osteoporosis was estimated using hazard ratio (HR) with 95% CI in Cox proportional hazard model, while mean level and trajectories of LDL-C for osteoporosis were evaluated using odds ratio (OR) with 95% CI in logistic regression model.
RESULTS:
During the median 6.2-year follow-up period, 70 men developed primary osteoporosis. The higher level of baseline LDL-C (HR = 1.539, 95% CI: 1.012-2.342) and mean LDL-C (OR = 2.190, 95% CI: 1.443-3.324) were associated with higher risk of osteoporosis in men with coronary heart disease after adjusted for covariates. Compared with those in the LDL-C trajectory of low-stable decrease, participants with medium-fluctuant trajectory, whose longitudinal LDL-C started with a medium LDL-C level and appeared an increase and then decrease, were negatively associated with osteoporosis risk (OR = 2.451, 95% CI: 1.152-5.216). And participants with initially high LDL-C level and then a rapid decrease demonstrated a tendency towards reduced risk (OR = 0.718, 95% CI: 0.212-2.437).
CONCLUSIONS
Elevated LDL-C level and its long-term fluctuation may increase the risk of primary osteoporosis in men. Early controlling a stable level of LDL-C is also essential for bone health.
9.Changing distribution and resistance profiles of common pathogens isolated from urine in the CHINET Antimicrobial Resistance Surveillance Program,2015-2021
Yanming LI ; Mingxiang ZOU ; Wen'en LIU ; Yang YANG ; Fupin HU ; Demei ZHU ; Yingchun XU ; Xiaojiang ZHANG ; Fengbo ZHANG ; Ping JI ; Yi XIE ; Mei KANG ; Chuanqing WANG ; Pan FU ; Yuanhong XU ; Ying HUANG ; Ziyong SUN ; Zhongju CHEN ; Yuxing NI ; Jingyong SUN ; Yunzhuo CHU ; Sufei TIAN ; Zhidong HU ; Jin LI ; Yunsong YU ; Jie LIN ; Bin SHAN ; Yan DU ; Sufang GUO ; Lianhua WEI ; Fengmei ZOU ; Hong ZHANG ; Chun WANG ; Yunjian HU ; Xiaoman AI ; Chao ZHUO ; Danhong SU ; Dawen GUO ; Jinying ZHAO ; Hua YU ; Xiangning HUANG ; Yan JIN ; Chunhong SHAO ; Xuesong XU ; Chao YAN ; Shanmei WANG ; Yafei CHU ; Lixia ZHANG ; Juan MA ; Shuping ZHOU ; Yan ZHOU ; Lei ZHU ; Jinhua MENG ; Fang DONG ; Zhiyong LÜ ; Fangfang HU ; Han SHEN ; Wanqing ZHOU ; Wei JIA ; Gang LI ; Jinsong WU ; Yuemei LU ; Jihong LI ; Jinju DUAN ; Jianbang KANG ; Xiaobo MA ; Yanping ZHENG ; Ruyi GUO ; Yan ZHU ; Yunsheng CHEN ; Qing MENG ; Shifu WANG ; Xuefei HU ; Jilu SHEN ; Ruizhong WANG ; Hua FANG ; Bixia YU ; Yong ZHAO ; Ping GONG ; Kaizhen WENG ; Yirong ZHANG ; Jiangshan LIU ; Longfeng LIAO ; Hongqin GU ; Lin JIANG ; Wen HE ; Shunhong XUE ; Jiao FENG ; Chunlei YUE
Chinese Journal of Infection and Chemotherapy 2024;24(3):287-299
Objective To investigate the distribution and antimicrobial resistance profiles of the common pathogens isolated from urine from 2015 to 2021 in the CHINET Antimicrobial Resistance Surveillance Program.Methods The bacterial strains were isolated from urine and identified routinely in 51 hospitals across China in the CHINET Antimicrobial Resistance Surveillance Program from 2015 to 2021.Antimicrobial susceptibility was determined by Kirby-Bauer method,automatic microbiological analysis system and E-test according to the unified protocol.Results A total of 261 893 nonduplicate strains were isolated from urine specimen from 2015 to 2021,of which gram-positive bacteria accounted for 23.8%(62 219/261 893),and gram-negative bacteria 76.2%(199 674/261 893).The most common species were E.coli(46.7%),E.faecium(10.4%),K.pneumoniae(9.8%),E.faecalis(8.7%),P.mirabilis(3.5%),P.aeruginosa(3.4%),SS.agalactiae(2.6%),and E.cloacae(2.1%).The strains were more frequently isolated from inpatients versus outpatients and emergency patients,from females versus males,and from adults versus children.The prevalence of ESBLs-producing strains in E.coli,K.pneumoniae and P.mirabilis was 53.2%,52.8%and 37.0%,respectively.The prevalence of carbapenem-resistant strains in E.coli,K.pneumoniae,P.aeruginosa and A.baumannii was 1.7%,18.5%,16.4%,and 40.3%,respectively.Lower than 10%of the E.faecalis isolates were resistant to ampicillin,nitrofurantoin,linezolid,vancomycin,teicoplanin and fosfomycin.More than 90%of the E.faecium isolates were ressitant to ampicillin,levofloxacin and erythromycin.The percentage of strains resistant to vancomycin,linezolid or teicoplanin was<2%.The E.coli,K.pneumoniae,P.aeruginosa and A.baumannii strains isolated from ICU inpatients showed significantly higher resistance rates than the corresponding strains isolated from outpatients and non-ICU inpatients.Conclusions E.coli,Enterococcus and K.pneumoniae are the most common pathogens in urinary tract infection.The bacterial species and antimicrobial resistance of urinary isolates vary with different populations.More attention should be paid to antimicrobial resistance surveillance and reduce the irrational use of antimicrobial agents.
10.Changing resistance profiles of Enterococcus in hospitals across China:results from the CHINET Antimicrobial Resistance Surveillance Program,2015-2021
Na CHEN ; Ping JI ; Yang YANG ; Fupin HU ; Demei ZHU ; Yingchun XU ; Xiaojiang ZHANG ; Yi XIE ; Mei KANG ; Chuanqing WANG ; Pan FU ; Yuanhong XU ; Ying HUANG ; Ziyong SUN ; Zhongju CHEN ; Yuxing NI ; Jingyong SUN ; Yunzhuo CHU ; Sufei TIAN ; Zhidong HU ; Jin LI ; Yunsong YU ; Jie LIN ; Bin SHAN ; Yan DU ; Sufang GUO ; Lianhua WEI ; Fengmei ZOU ; Hong ZHANG ; Chun WANG ; Yunjian HU ; Xiaoman AI ; Chao ZHUO ; Danhong SU ; Dawen GUO ; Jinying ZHAO ; Hua YU ; Xiangning HUANG ; Wen'en LIU ; Yanming LI ; Yan JIN ; Chunhong SHAO ; Xuesong XU ; Chao YAN ; Shanmei WANG ; Yafei CHU ; Lixia ZHANG ; Juan MA ; Shuping ZHOU ; Yan ZHOU ; Lei ZHU ; Jinhua MENG ; Fang DONG ; Zhiyong LÜ ; Fangfang HU ; Han SHEN ; Wanqing ZHOU ; Wei JIA ; Gang LI ; Jinsong WU ; Yuemei LU ; Jihong LI ; Jinju DUAN ; Jianbang KANG ; Xiaobo MA ; Yanping ZHENG ; Ruyi GUO ; Yan ZHU ; Yunsheng CHEN ; Qing MENG ; Shifu WANG ; Xuefei HU ; Jilu SHEN ; Ruizhong WANG ; Hua FANG ; Bixia YU ; Yong ZHAO ; Ping GONG ; Kaizhen WEN ; Yirong ZHANG ; Jiangshan LIU ; Longfeng LIAO ; Hongqin GU ; Lin JIANG ; Wen HE ; Shunhong XUE ; Jiao FENG ; Chunlei YUE
Chinese Journal of Infection and Chemotherapy 2024;24(3):300-308
Objective To understand the distribution and changing resistance profiles of clinical isolates of Enterococcus in hospitals across China from 2015 to 2021.Methods Antimicrobial susceptibility testing was conducted for the clinical isolates of Enterococcus according to the unified protocol of CHINET program by automated systems,Kirby-Bauer method,or E-test strip.The results were interpreted according to the Clinical & Laboratory Standards Institute(CLSI)breakpoints in 2021.WHONET 5.6 software was used for statistical analysis.Results A total of 124 565 strains of Enterococcus were isolated during the 7-year period,mainly including Enterococcus faecalis(50.7%)and Enterococcus faecalis(41.5%).The strains were mainly isolated from urinary tract specimens(46.9%±2.6%),and primarily from the patients in the department of internal medicine,surgery and ICU.E.faecium and E.faecalis strains showed low level resistance rate to vancomycin,teicoplanin and linezolid(≤3.6%).The prevalence of vancomycin-resistant E.faecalis and E.faecium was 0.1%and 1.3%,respectively.The prevalence of linezolid-resistant E.faecalis increased from 0.7%in 2015 to 3.4%in 2021,while the prevalence of linezolid-resistant E.faecium was 0.3%.Conclusions The clinical isolates of Enterococcus were still highly susceptible to vancomycin,teicoplanin,and linezolid,evidenced by a low resistance rate.However,the prevalence of linezolid-resistant E.faecalis was increasing during the 7-year period.It is necessary to strengthen antimicrobial resistance surveillance to effectively identify the emergence of antibiotic-resistant bacteria and curb the spread of resistant pathogens.

Result Analysis
Print
Save
E-mail