1.Explainable Machine Learning Model for Predicting Prognosis in Patients with Malignant Tumors Complicated by Acute Respiratory Failure: Based on the eICU Collaborative Research Database in the United States
Zihan NAN ; Linan HAN ; Suwei LI ; Ziyi ZHU ; Qinqin ZHU ; Yan DUAN ; Xiaoting WANG ; Lixia LIU
Medical Journal of Peking Union Medical College Hospital 2026;17(1):98-108
To develop and validate a model for predicting intensive care unit (ICU) mortality risk in patients with malignant tumors complicated by acute respiratory failure (ARF) based on an explainable machine learning framework. Clinical data of patients with malignant tumors and ARF were extracted from the eICU Collaborative Research Database in the United States, including demographic characteristics, comorbidities, vital signs, laboratory test indicators, and major interventions within the first 24 hours after ICU admission.The study outcome was ICU death.Enrolled patients were randomly divided into a training set and a validation set at a ratio of 7:3.Predictor variables were selected using least absolute shrinkage and selection operator (LASSO) regression.Five machine learning algorithms-extreme gradient boosting (XGBoost), support vector machine (SVM), Logistic regression, multilayer perceptron (MLP), and C5.0 Decision Tree-were employed to construct predictive models.Model performance was evaluated based on the area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, and other metrics.The optimal model was further interpreted using the Shapley additive explanations (SHAP) algorithm. A total of 3196 patients with malignant tumors complicated by ARF were included.The training set comprised 2, 261 patients and the validation set 935 patients; 683 patients died during ICU stay, while 2513 survived.LASSO regression ultimately selected 12 variables closely associated with patient ICU outcomes, including sepsis comorbidity, use of vasoactive drugs, and within the first 24 hours after ICU admission: minimum mean arterial pressure, maximum heart rate, maximum respiratory rate, minimum oxygen saturation, minimum serum bicarbonate, minimum blood urea nitrogen, maximum white blood cell count, maximum mean corpuscular volume, maximum serum potassium, and maximum blood glucose.After model evaluation, the XGBoost model demonstrated the best performance.The AUCs for predicting ICU mortality risk in the training and validation sets were 0.940 and 0.763, respectively; accuracy was 88.3% and 81.2%;sensitivity was 98.5% and 95.9%.Its predictive performance also remained optimal in sensitivity analyses.SHAP analysis indicated that the top five variables contributing to the model's predictions were minimum oxygen saturation, minimum serum bicarbonate, minimum mean arterial pressure, use of vasoactive drugs, and maximum white blood cell count. This study successfully developed a mortality risk prediction model for ICU patients with malignant tumors complicated by ARF based on a large-scale dataset and performed explainability analysis.The model aids clinicians in early identification of high-risk patients and implementing individualized interventions.
2.Association between urinary levels of six per- and poly-fluoroalkyl substances in early pregnancy and risk of gestational diabetes mellitus
Ziyi LIU ; Luming YAN ; Tingting JIANG ; Yaling LI ; Chao ZHANG ; Jiahu HAO
Journal of Environmental and Occupational Medicine 2026;43(2):174-181
Background Per- and poly-fluoroalkyl substances (PFAS) can influence gestational diabetes mellitus (GDM); however, current studies on their association are limited and have yielded inconsistent findings. Objective To investigate the association between maternal exposure to PFAS, as measured by urinary concentrations in early pregnancy, and the risk of developing GDM. Methods Based on the Wuhu Birth Cohort in Anhui Province conducted between 2020 and 2023, this study included
3.Association of latent profiles of mobile phone dependence and self control with physical exercise among junior high school students
YAN Chuqi, ZHANG Tiancheng, ZHANG Fulan, WANG Aolun, PENG Jinyin, CHEN Ziyi, LUO Xiangyu
Chinese Journal of School Health 2025;46(3):391-395
Objective:
To explore the association of latent profiles of mobile phone dependence and self control with physical exercise among junior high school students, so as to provide references for the prevention of mobile phone dependence and the improvement of self control among junior high school students.
Methods:
From April to May 2024, a stratified random cluster sampling method was used to select a total of 2 311 students from grade 7 to grade 9 in three public junior high schools in Xiangxi Autonomous Prefecture, Hunan Province. Latent profile analysis was conducted to identify the latent profiles of mobile phone dependence and self control among junior high school students. Pearson correlation analysis was used to examine the correlation between mobile phone dependence and self control, and Chi square test was used to analyze the distribution differences of latent profiles of adolescents across different demographic characteristics. Multiple Logistic regression analysis was applied to explore the association between mobile phone dependence, self control, and physical exercise.
Results:
Four latent profiles of mobile phone dependence and self control were identified: low dependence-moderate self control group ( n =885, 38.3%), moderate dependence-low self control group ( n =910, 39.4%), high dependence-no self control group ( n =232, 10.0%), and no dependence-high self control group ( n =284, 12.3%). Significant differences were observed in the distribution of latent profiles across gender, grade and only child status ( χ 2=10.85, 35.72, 13.85, P <0.05). Logistic regression analysis showed that, after controlling for demographic variables, compared with the low dependence-moderate self control group, physical exercise was negatively associated with the moderate dependence-low self control group ( OR =0.79) and the high dependence-no self control group ( OR =0.81), while positively associated with the no dependence-high self control group ( OR =1.58) ( P <0.01).
Conclusions
The influence of physical exercise on junior high school students different potential profile types of mobile phone dependence and self control is different. Schools and families should adopt targeted physical exercise interventions based on the characteristics of different profiles to promote the physical and mental health of junior high school students.
4.Development trajectory of mobile phone dependence in middle school students and its association with loneliness and self-control
LUO Xiangyu, ZHANG Tiancheng, WANG Aolun, ZHANG Fulan, LIU Yang, YAN Chuqi, CHEN Ziyi
Chinese Journal of School Health 2025;46(5):624-629
Objective:
To analyze the heterogeneity of mobile phone dependence development trajectory in middle school students and its association with loneliness and selfcontrol ability, so as to provide reference for the prevention of mobile phone dependence in middle school students.
Methods:
A total of 941 grade 1 students from 4 public middle schools in Xiangxi Autonomous Prefecture, Hunan Province were selected for the followup survey by random cluster sampling from October 2023 to April 2024 and October 2024. Mobile Phone Addiction Index (MPAI), University of California, Los Angeles Loneliness Scale-20 (UCLA-20) and Selfcontrol Scales (SCS) were used for questionnaire survey. The heterogeneity of the developmental trajectory of middle school students mobile phone dependence was analyzed by the latent growth curve model (LGMM), and the influencing factors of the developmental trajectory of middle school students mobile phone dependence were explored by multiple Logistic regression analysis.
Results:
The development trajectory of middle school students mobile phone dependence could be divided into four categories: C1 "low risk slow decline group (n=438,44.6%)", C2 "medium risk slow rise group (n=272,29.7%)", C3 "high risk rapid decline group (n=73,8.6%)" and C4 "high risk rapid rise group (n=158,17.1%)". There were significant differences in the distribution of mobile phone dependence development track heterogeneity subgroups among sex, only child, lodging, and leftbehind students (χ2=117.79, 44.88, 37.09, 130.50, P <0.01). The results of the multinomial Logistic regression model analysis showed that, with C1 group as the reference, C2, C3, and C4 were positively correlated with students loneliness [OR(95%CI)=1.04 (1.02-1.06), 1.11(1.08-1.14), 1.12(1.09-1.14)]; C2 and C4 groups were negatively correlated with students selfcontrol [OR(95%CI)=0.97(0.96-0.99), 0.95(0.93-0.97)] (P<0.01).
Conclusions
The development trajectory of mobile phone dependence among middle school students is heterogeneous. Reducing the loneliness of individuals and cultivating good selfcontrol ability are helpful to alleviate mobile phone dependence behavior among middle school students.
5.Periodontitis Aggravates Chronic Obstructive Pulmonary Disease Progression by Promoting Pulmonary Macrophage M1 Polarizations
Jing LI ; Kaixin XIONG ; Boyu TANG ; Ziyi XIA ; Zhao TANG ; Yan LI
Journal of Sichuan University (Medical Sciences) 2025;56(5):1281-1288
Objective To investigate the critical role of macrophage M1 polarization in mediating the effect of periodontitis on the progression of chronic obstructive pulmonary disease(COPD).Methods Alveolar lavage fluid samples were collected from COPD patients with comorbid periodontitis,and gene expression analysis was performed to validate the changes in the expression of M1 polarization-related genes.A mouse model of COPD,with experimentally induced periodontitis,were established.Hematoxylin and eosin(HE)staining of pathological sections was performed to observe the effect of periodontitis on COPD progression.Flow cytometry,immunofluorescence staining,and reverse transcription quantitative polymerase chain reaction(RT-qPCR)were performed to analyze the effect of periodontitis on macrophage M1 polarization and the expression of relevant genes in the alveolar lavage fluid and lung tissues.Results In clinical samples of alveolar lavage fluid from COPD patients with periodontitis,the expression of macrophage M1 polarization-related genes,including CD86,inducible nitric oxide synthase(iNOS),interleukin(IL)-1β,tumor necrosis factor(TNF)-α,IL-23,and IL-6,was upregulated compared with that of COPD patients without periodontitis.Analysis of a mouse disease model revealed that periodontitis affected the growth of COPD mice,with the final body mass of mice in the periodontitis and COPD comorbid group([21.3±0.52]g,day 34)lower than that of the COPD group([23.93±0.45]g,day 34).Pathological sections of the lung tissue showed that periodontitis exacerbated COPD progression,with more pronounced alveolar expansion and alveolar wall destruction observed in the periodontitis and COPD comorbid group.Flow cytometry revealed a higher proportion of M1-polarized macrophages in alveolar lavage fluid from COPD and periodontitis comorbid mice([31.36±2.51]%)compared with the COPD mice([23.19±1.07]%).Immunofluorescence assays indicated that periodontitis also promoted macrophage M1 polarization in the lung tissue of COPD mice.Gene expression analysis demonstrated that M1 polarization-related gene expression was significantly upregulated in both the alveolar lavage fluid and lung tissue of mice in the COPD and periodontitis co-morbid group compared to the COPD group.Conclusion Periodontitis exacerbates COPD progression by promoting macrophage M1 polarization in the lungs.Enhancing oral hygiene management and targeting the inhibition of macrophage M1 polarization may represent new therapeutic strategies for the clinical prevention and control of COPD.
6.Chaihu Shugansan Combined with Ferulic Acid Regulates BDNF/TrkB Signaling Pathway and Monoamine Neurotransmitters in Frontal Cortex of Rat Model of CUMS
Yuexin LI ; Zhijing ZHANG ; Ziyi GUO ; Di YAN ; Xueyan HU ; Jianping YAO
Chinese Journal of Experimental Traditional Medical Formulae 2025;31(15):28-37
ObjectiveTo observe the antidepressant effect of Chaihu Shugansan combined with ferulic acid on the rat model of chronic unpredictable mild stress (CUMS) and explore the mechanism from the histomorphology of frontal cortex, expression of key molecules in the brain-derived neurotrophic factor (BDNF)/tyrosine kinase receptor B (TrkB) signaling pathway, and changes in monoamine neurotransmitter levels. MethodsSixty adult male SD rats were randomized into six groups (n=10): blank control, depression model, Chaihu Shugansan (3.3 g·kg-1·d-1), ferulic acid (50 mg·kg-1·d-1), Chaihu Shugansan (3.3 g·kg-1·d-1) + ferulic acid (50 mg·kg-1·d-1), and fluoxetine (2.1 mg·kg-1·d-1). Rats in other groups except the blank control group were subjected to a mild chronic unpredictable stress stimulus every day. Seven stimuli were used, including fasting with free access to water for 24 h, water deprivation with free access to food for 24 h, wetting the bedding with water in the cage, restraint for 3 h, tail clamping for 1 min, swimming in ice water at 4 ℃, and day and night reversal. Each stimulus was used 1 to 3 times, and the modeling lasted for a total of 21 days. At the same time of stimulation, rats in each medication group were treated with corresponding agents by gavage, while those in the blank control group and the depression model group received equal volumes of normal saline by gavage. The open field test, sucrose preference test, and forced swimming test were conducted before and after modeling. The rats were anesthetized by intraperitoneal injection of 3% pentobarbital sodium, and the frontal cortex was isolated on ice. The mRNA and protein levels of BDNF, TrkB, and cyclic adenosine monophosphate-responsive element-binding protein (CREB) in the frontal cortex were determined by Real-time fluorescence quantitative polymerase chain reaction (Real-time PCR) and Western blot, respectively. The levels of monoamine neurotransmitters 5-hydroxytryptamine (5-HT), dopamine (DA), and norepinephrine (NE) in the frontal cortex were determined by enzyme-linked immunosorbent assay. Light microscopy was employed to observe the histopathological changes in the frontal cortex. ResultsCompared with the blank control group, the depression model group showed reduced body mass (P<0.05, P<0.01), decreased number of crossings and rearings in the open field test and sucrose preference (P<0.01), prolonged time of immobility in the forced swimming test (P<0.01), reduced neuronal cells, increased necrotic cells, and darkening cell staining in the frontal cortex, down-regulated mRNA and protein levels of BDNF, TrkB, CREB, and lowered levels of 5-HT, NE, and DA in the frontal cortex (P<0.01). Compared with the depression model group, each intervention group showed improved general state, increased body mass (P<0.05), increased number of crossings (P<0.05), shortened immobility time in the forced swimming test (P<0.01), increased neuronal cells, reduced necrotic cells, and lightened cellular staining in the frontal cortex, up-regulated mRNA and protein levels of BDNF, TrkB and CREB, and elevated levels of 5-HT, NE, and DA in the frontal cortex (P<0.01). Moreover, the Chaihu Shugansan + ferulic acid group outperformed the Chaihu Shugansan group and the ferulic acid group in increasing the body mass and the 5-HT content in the frontal cortex (P<0.05). The combination group outperformed the Chaihu Shugansan group regarding the number of rearings and up-regulation in the mRNA level of BDNF in the frontal cortex (P<0.05), and it was superior to the ferulic acid group in terms of shortening the immobility time in the forced swimming test, up-regulating the mRNA levels of BDNF, TrkB, and CREB and the protein levels of BDNF and CREB in the frontal cortex, and increasing the DA content in the frontal cortex (P<0.05). ConclusionChaihu Shugansan combined with ferulic acid can exert antidepressant effect on the rat model of CUMS by regulating the BDNF/TrkB signaling pathway and monoamine neurotransmitter content in the frontal cortex. Moreover, the antidepressant effect of Chaihu Shugansan combined with ferulic acid was more significant than that of Chaihu Shugansan and ferulic acid used alone.
7.An interpretable machine learning modeling method for the effect of manual acupuncture manipulations on subcutaneous muscle tissue.
Wenqi ZHANG ; Yanan ZHANG ; Yan SHEN ; Chun SUN ; Jie CHEN ; Yuhe WEI ; Jian KANG ; Ziyi CHEN ; Jingqi YANG ; Jingwen YANG ; Chong SU
Chinese Acupuncture & Moxibustion 2025;45(10):1371-1382
OBJECTIVE:
To investigate the effect of manual acupuncture manipulations (MAMs) on subcutaneous muscle tissue, by developing quantitative models of "lifting and thrusting" and "twisting and rotating", based on machine learning techniques.
METHODS:
A depth camera was used to capture the acupuncture operator's hand movements during "lifting and thrusting" and "twisting and rotating" of needle. Simultaneously, the ultrasound imaging was employed to record the muscle tissue responses of the participants. Amplitude and angular features were extracted from the movement data of operators, and muscle fascicle slope features were derived from the data of ultrasound images. The dynamic time warping barycenter averaging algorithm was adopted to align the dual-source data. Various machine learning techniques were applied to build quantitative models, and the performance of each model was compared. The most optimal model was further analyzed for its interpretability.
RESULTS:
Among the quantitative models built for the two types of MAMs, the random forest model demonstrated the best performance. For the quantitative model of the "lifting and thrusting" technique, the coefficient of determination (R2) was 0.825. For the "twisting and rotating" technique, R2 reached 0.872.
CONCLUSION
Machine learning can be used to effectively develop the models and quantify the effects of MAMs on subcutaneous muscle tissue. It provides a new perspective to understand the mechanism of acupuncture therapy and lays a foundation for optimizing acupuncture technology and designing personalized treatment regimen in the future.
Humans
;
Acupuncture Therapy/methods*
;
Machine Learning
;
Male
;
Adult
;
Female
;
Subcutaneous Tissue/diagnostic imaging*
;
Young Adult
8.Establishment of the epidemiological cut-off value for antifungal drugs against Cryptococcus neoformans in East China from 2017 to 2022
Lili WANG ; Ziyi ZHOU ; Ziwen WANG ; Wenjie TIAN ; Yan ZHANG ; Lingbing ZENG ; Wenjuan WU
Chinese Journal of Laboratory Medicine 2024;47(6):639-643
Objective:To establish an epidemiological cut-off value (ECV) for antifungal drugs against Cryptococcus neoformans in East China through a multicenter in vitro drug susceptibility test. Methods:A retrospective collection of 479 clinical isolates of Cryptococcus neoformans was conducted by the East China Invasive Fungal Infection Group (ECIFIG) from January 1, 2017 to December 31, 2022. Mass spectrometry and gene sequencing were used for identification. A unified drug susceptibility testing system was established in the fungal laboratories across three sub centers in Shanghai, Jiangxi, and Jiangsu provinces. Drug susceptibility testings of Cryptococcus neoformans were independently completed in each center after passing consistency evaluation. Epidemiological breakpoints were established against fluconazole, voriconazole, amphotericin B, 5-fluorocytosine, isaconazole, posaconazole and itraconazole following the principles and procedures of the Clinical and Laboratory Standards Institute (CLSI) M57. Results:External consistency evaluation revealed that the minimum inhibitory concentration values ??for all drugs in each center did not differ by more than one dilution gradient. A new ECV type of Cryptococcus neoformans in East China was established, including fluconazole 16 mg/L, voriconazole 0.12 mg/L, amphotericin B 1 mg/L, 5-fluorocytosine 8 mg/L, isavuconazole 0.12 mg/L, posaconazole 0.5 mg/L, and itraconazole 0.5 mg/L. Conclusion:This study preliminarily revealed the drug susceptibility characteristics of clinical Cryptococcus neoformans in East China and established the ECVs for antifungal drugs against Cryptococcus neoformans in the region.
9.Diagnostic Efficacy of Platelet-Related Parameters on Anxiety and Depression in Patients Undergoing Peritoneal Dialysis
Chenling LIU ; Jingyi ZHU ; Linlin WANG ; Yuan GAO ; Ziyi YAN ; Jiayin WANG ; Shengjun LIU
Acta Academiae Medicinae Sinicae 2024;46(1):43-48
Objective To analyze the correlations between platelet-related parameters and the incidence of anxiety and depression in the patients undergoing peritoneal dialysis(PD),and evaluate the efficacy of the pa-rameters in the diagnosis of anxiety and depression in PD patients.Methods A total of 245 patients undergoing PD in the First Affiliated Hospital of Hebei North University from September 2022 to February 2023 were enrolled.The gener-alized anxiety scale(GAD-7)and the patient health questionnaire(PHQ-9)were used to evaluate the anxiety and depression of the patients,respectively.The personal information and biochemical indicators of the patients were col-lected,and the platelet count(PLT),mean platelet volume(MPV),and platelet distribution width(PDW)were measured.Logistic regression was adopted to analyze the relationships of platelet-related parameters with anxiety and depression in PD patients.Results Among the 245 patients undergoing PD,the incidences of anxiety and depression were 15.9% and 38.0% ,respectively.There were differences in the dialysis period(Z=-2.358,P =0.018;Z =-3.079,P=0.002),MPV(Z=-4.953,P<0.001;Z=-7.878,P<0.001),and PDW(Z =-4.587,P<0.001;Z=-7.367,P<0.001)between the anxiety group and the non-anxiety group as well as between the de-pression group and the non-depression group.The correlation analysis showed that MPV(r =0.358,P<0.001;r =0.489,P<0.001)and PDW(r =0.340,P<0.001;r =0.447,P<0.001)were positively correlated with anxiety and depression in the patients undergoing PD.The Logistic regression model showed that MPV(P =0.022,P =0.011),PDW(P =0.041,P =0.018),and dialysis period(P =0.011,P =0.030)were independent risk factors for the anxiety and depressive state in PD patients.The areas under the receiver operating characteristic curve of MPV in the diagnosis of anxiety and depression in PD patients were 0.750 and 0.800,respectively,and those of PDW were 0.732 and 0.780,respectively.Conclusion MPV and PDW have high efficacy in the diagnosis of anxiety and depression associated with PD and can be used as objective indicators to evaluate the anxiety and depression in the patients undergoing PD.
10.Analytic method of the characteristics of acupuncture manipulation based on ultrasound imaging
Jie CHEN ; Jun ZHAO ; Yuhe WEI ; Yang BAI ; Jiyu HE ; Ziyi CHEN ; Liming SUN ; Lei WANG ; Jingli LI ; Yanan ZHANG ; Yan SHEN ; Chong SU
China Medical Equipment 2024;21(10):10-18
Objective:To construct an analytic method aimed at the characteristics of the commonly method of supplementing and pouring of acupuncture based on the analysis and modeling of ultrasound images around acupoint region in the process of acupuncture.Methods:A total of 7 healthy subjects who underwent physical examination in Beijing Zhongguancun Hospital from June,2022 to June,2023 were selected,and their Kongzui acupoints were acupunctured by 10 acupuncturists with associate senior title as 4 kinds of acupuncture manipulations included reinforcing by twisting and rotating(RFTR),reducing by twisting and rotating(RDTR),reinforcing by lifting and thrusting(RFLT),and reducing by lifting and thrusting(RDLT).The B-ultrasound diagnostic device was used to collect the images of muscle and fascial tissue below the acupoint,so as to construct the model of images.The definition of virtual acupuncture point was adopted to study the regulation of perturbation of subcutaneous tissue that was caused after the skin was acupunctured by needle.The change regulation of the virtual acupuncture point of muscle bundle below skin at Zuikong acupoint of subjects was analyzed.Results:The difference value of average absolution value between peak and trough of the trajectory of virtual acupuncture point of twisting and rotating was 0.066±0.045,and the average value of amplitude of this method was less than that(0.428±0.276)of lifting and thrusting method,and the twisting and rotating method was uniform and symmetrical,and there was difference between two kinds of acupuncture methods.The characteristics of computer graphics was used to qualify the work effect of lifting and thrusting,and reinforcing and reducing,which showed the heavy insertion and light lifting of RFLT,and showed heavy lifting and light insertion of RDLT,thus distinguished the two methods[(RFLT)and(RDLT)].Conclusions:The ultrasound imaging and computer graphics can be used to analyze the regularity of the common"reinforcing and reducing"method of acupuncture and moxibustion.


Result Analysis
Print
Save
E-mail