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.Analysis of unexamined items in physical examinations of radiation workers at an occupational health examination institution in Henan Province, China, 2023
Lihong MA ; Fengling ZHAO ; Yuzheng LI ; Han LIU ; Yumin LV
Chinese Journal of Radiological Health 2026;35(1):12-17
Objective To analyze the unexamined items and situations in occupational health examinations of radiation workers, and provide a reference for the revision of occupational health examination standards for radiation workers. Methods A total of 29 630 radiation workers who underwent occupational health examinations at The Third People’s Hospital of Henan Province in 2023 were selected, and the non-examination rates were statistically analyzed according to occupation, gender, and age. Results The overall non-examination rate of non-medical radiation workers was significantly lower than that of the medical radiation workers (P<0.05). The non-examination rate of chest X-rays among medical radiation workers was significantly higher than that of non-medical radiation workers (P<0.05), while no significant differences were found in other items (P>0.05). Gender-stratified analysis showed that the non-examination rate of routine urine tests was higher in females than in males in both medical and non-medical radiation workers (P<0.05). Age-stratified analysis revealed no significant differences in non-examination rates among different age groups in non-medical radiation workers (P>0.05), whereas the chest X-ray non-examination rate was relatively high in medical radiation workers under 30 years old (P<0.05). Conclusion Significant differences were observed in the non-examination rates of occupational health examinations among radiation workers based on occupation, gender, and age. The overall non-examination rate was relatively low in non-medical radiation workers.
3.Relationship between ischemia-reperfusion injury and rejection
Aizheng XI ; Han XU ; Jun LUO ; Zhongzhong LIU
Organ Transplantation 2026;17(2):304-310
Ischemia-reperfusion injury of solid organs is a common complication during transplantation, but its mechanism remains unclear and there are no effective prevention and treatment methods. Rejection is a common immune response of the graft, which is classified into hyperacute rejection, acute rejection and chronic rejection. Rejection and ischemia-reperfusion injury of solid organs may share some common mechanisms, and further research on their correlation may provide strategies for the prevention and treatment of ischemia-reperfusion injury of solid organs. This article mainly reviews the association between rejection and ischemia-reperfusion injury of solid organs.
4.Flavonoids Intervene in Diabetic Nephropathy by Regulating TGF-β/Smad Signaling Pathway: A Review
Qihui QIU ; Chang LIU ; Xiaotong YAN ; Jinwei HAN ; Hui SUN ; Fengting YIN ; Yuhang WANG ; Mengmeng WANG ; Xijun WANG
Chinese Journal of Experimental Traditional Medical Formulae 2026;32(7):300-309
Diabetic nephropathy (DKD), as a common microvascular complication of diabetes mellitus (DM), is a major cause of end-stage renal disease (ESRD). Its clinical manifestations include increased urinary protein excretion, thickening of the glomerular basement membrane, and renal tubulointerstitial fibrosis. The pathogenesis of DKD is complex and involves multiple factors, including disordered glucose metabolism, hemodynamic alterations, and oxidative stress. Although modern medical approaches can alleviate certain symptoms, they still have limitations such as insufficient therapeutic targeting and prominent adverse effects. The transforming growth factor-β/Smad (TGF-β/Smad) signaling pathway is not only a tissue fibrosis pathway that has attracted considerable attention in recent years, but also regulates multiple protein molecules, including the glomerular podocyte slit diaphragm protein Podocin, interleukin-1β (IL-1β), and superoxide dismutase (SOD), thereby participating in various pathological processes and ultimately mediating renal injury. Flavonoid compounds, owing to their sustained pharmacological effects, broad spectrum of action, and high safety profile, have become ideal candidates for targeted therapy research in DKD. Existing studies have shown that these compounds can exert inhibitory effects on renal fibrosis, alleviate inflammatory responses, protect podocytes, and reduce oxidative stress by regulating the interactions between the TGF-β/Smad signaling pathway and the aforementioned protein molecules, thereby maintaining renal structure and function, reducing proteinuria, and significantly improving DKD lesions. This review briefly outlines the composition and functions of the TGF-β/Smad signaling pathway, elucidates the mechanisms by which this pathway regulates DKD, and focuses on summarizing major studies from the past decade on flavonoid-based interventions in DKD through targeted inhibition of the TGF-β/Smad signaling pathway. Furthermore, it discusses the considerable therapeutic potential of flavonoids in the treatment of this disease, aiming to provide a scientific basis for future clinical prevention and treatment of DKD and to promote the development of targeted drugs.
5.QingNangTCM: a parameter-efficient fine-tuning large language model for traditional Chinese medicine
Xuming TONG ; Liyan LIU ; Yanhong YUAN ; Xiaozheng DING ; Huiru JIA ; Xu YANG ; Sio Kei IM ; Mini Han WANG ; Zhang XIONH ; Yapeng WANG
Digital Chinese Medicine 2026;9(1):1-12
Objective:
To develop QingNangTCM, a specialized large language model (LLM) tailored for expert-level traditional Chinese medicine (TCM) question-answering and clinical reasoning, addressing the scarcity of domain-specific corpora and specialized alignment.
Methods:
We constructed QnTCM_Dataset, a corpus of 100 000 entries, by integrating data from ShenNong_TCM_Dataset and SymMap v2.0, and synthesizing additional samples via retrieval-augmented generation (RAG) and persona-driven generation. The dataset comprehensively covers diagnostic inquiries, prescriptions, and herbal knowledge. Utilizing P-Tuning v2, we fine-tuned the GLM-4-9B-Chat backbone to develop QingNangTCM. A multi-dimensional evaluation framework, assessing accuracy, coverage, consistency, safety, professionalism, and fluency, was established using metrics such as bilingual evaluation understudy (BLEU), recall-oriented understudy for gisting evaluation (ROUGE), metric for evaluation of translation with explicit ordering (METEOR), and LLM-as-a-Judge with expert review. Qualitative analysis was conducted across four simulated clinical scenarios: symptom analysis, disease treatment, herb inquiry, and failure cases. Baseline models included GLM-4-9B-Chat, DeepSeek-V2, HuatuoGPT-II (7B), and GLM-4-9B-Chat (freeze-tuning).
Results:
QingNangTCM achieved the highest scores in BLEU-1/2/3/4 (0.425/0.298/0.137/0.064), ROUGE-1/2 (0.368/0.157), and METEOR (0.218), demonstrating a balanced and superior normalized performance profile of 0.900 across the dimensions of accuracy, coverage, and consistency. Although its ROUGE-L score (0.299) was lower than that of HuatuoGPT-II (7B) (0.351), it significantly outperformed domain-specific models in expert-validated win rates for professionalism (86%) and safety (73%). Qualitative analysis confirmed that the model strictly adheres to the “symptom-syndrome-pathogenesis-treatment” reasoning chain, though occasional misclassifications and hallucinations persisted when dealing with rare medicinal materials and uncommon syndromes.
Conclusion
Combining domain-specific corpus construction with parameter-efficient prompt tuning enhances the reasoning behavior and domain adaptation of LLMs for TCM-related tasks. This work provides a technical framework for the digital organization and intelligent utilization of TCM knowledge, with potential value for supporting diagnostic reasoning and medical education.
6.The feasibility of using high-definition thoracoscopy to identify sympathetic ganglia during thoracic sympathicotomy for primary palmar hyperhidrosis
Gang XU ; Chaoyue HU ; Cong CHEN ; Yuancai LIN ; Daolong ZHU ; Han LIU ; Dong WANG
Chinese Journal of Clinical Thoracic and Cardiovascular Surgery 2026;33(04):578-583
Objective To explore the feasibility of using high-definition thoracoscopy to identify sympathetic ganglia during thoracic sympathicotomy for primary palmar hyperhidrosis. Methods The clinical data of patients with primary palmar hyperhidrosis who underwent high-definition thoracoscopic sympathicotomy in Taikang Xianlin Drum Tower Hospital from June to July 2023 were retrospectively analyzed. Intraoperative visualization rates and anatomical variations of sympathetic ganglia were recorded, and the consistency between white-light thoracoscopy and near-infrared fluorescence imaging was compared. Additionally, surgical videos from previous fluorescence-guided procedures were reviewed. Results Finally 100 patients were collected, including 54 females and 46 males, with an average age of (21.92±6.56) years. All patients underwent endoscopic thoracic sympathicotomy at R3 level. The overall intraoperative ganglion visualization rate was 92.5% (740/800), with G2-G5 rates of 95.5% (191/200), 94.0% (188/200), 94.0% (188/200), and 86.5% (173/200), respectively. Ganglion variations occurred in 32.0% (237/740), predominantly at G3 (29.8%) and G4 (42.6%). In 5 indocyanine green-enhanced patients, the concordance rate between white-light and near-infrared fluorescence imaging was 100.0% (38/38). Video analysis of 14 near-infrared fluorescence-guided surgeries demonstrated a 99.1% (107/108) consistency rate. Postoperative palmar hyperhidrosis improvement reached 100.0% (100/100) with no Horner’s syndrome. Conclusion With the wide clinical application of high-definition thoracoscopy, accurate thoracic sympathicotomy has the feasibility of clinical application.
7.Expert consensus on clinical application of parenteral direct thrombin inhibitors in perioperative period
Mingyu JIANG ; Yuan BIAN ; Lizhu HAN ; Qinan YIN ; Fengjiao KANG ; Anhua WEI ; Danjie ZHAO ; Lin WANG ; Ying SHAO ; Li TANG ; Yi WANG ; Shuhong LIANG ; Huijuan LIU ; Guirong XIAO ; Yue LI
China Pharmacy 2026;37(6):689-699
OBJECTIVE To form an expert consensus on the clinical application of parenteral direct thrombin inhibitors (DTIs) in patients during the perioperative period. METHODS Led by Sichuan Academy of Medical Sciences & Sichuan Provincial People’s Hospital (the Affiliated Hospital of UESTC), a multidisciplinary working group was established. Through literature review and the Delphi method, clinical questions related to the rational perioperative use of parenteral DTIs were identified. A structured design was adopted using the “Population-Intervention-Comparison-Outcome” framework; systematic searches were conducted in CNKI, Medline, Embase and other databases. Relevant evidence from randomized controlled trials and cohort studies was included and synthesized. Evidence quality was assessed using the Grades of Recommendations Assessment,Development and Evaluation (GRADE) approach, and recommendations were formulated through multiple rounds of Delphi surveys and expert consensus meetings. RESULTS &CONCLUSIONS Seven recommendations (each with an expert consensus rate exceeding 90%) on the use of parenteral DTIs in perioperative patients were developed. These recommendations specify drug selection, dosing ranges, key monitoring points, and safety management strategies for parenteral DTIs in various scenarios, including the perioperative period of ventricular assist device implantation, the perioperative period of cardiac surgery, perioperative patients with lower-extremity atherosclerotic disease, the perioperative period of percutaneous coronary intervention in patients with acute coronary syndrome, the perioperative period of carotid artery stenting in patients with carotid stenosis, the perioperative period of patients with right heart thrombosis, and patients who develop related thrombosis and dysfunction after a central venous catheter insertion. In addition, warning and management pathways for perioperative bleeding and thrombotic events were proposed. This expert consensus, which is formulated based on the best available evidence, provides evidence-based guidance for standardized and individualized use of parenteral DTIs in perioperative period.
8.Impact of rapid antiviral therapy on HIV/AIDS mortality outcomes in Wuhan
Jie DING ; Xuejiao HU ; Han YAN ; Rong HU ; Pulin LIU
Journal of Public Health and Preventive Medicine 2026;37(2):94-98
Objective To analyze the situation of rapid antiretroviral therapy (ART) and death of HIV/AIDS in Wuhan from 1994 to 2023, and to provide a scientific basis for further rapid initiation of ART and reduction of mortality rate. Methods According to the case follow-up and treatment database of China AIDS Prevention and Control Information System, data were obtained from all the cases reported from January 1, 1994 to December 31, 2023 with the current address in Wuhan City and the review status of the final review card. The data were analyzed using Kaplan-Meier and Cox proportional hazards models. Results The total mortality rate of HIV/AIDS in Wuhan from 1994 to 2023 was 12.76%. The proportion of receiving antiretroviral therapy within 30 days increased year by year, and the mortality rate decreased year by year. After adjusting for sex, age, occupation, ethnicity, education level, mobile population, history of STD, route of infection, source of sample, and first CD4 value, receiving antiretroviral therapy within 30 days (HR=0.08, 95%CI: 0.07-0.10) was a protective factor for HIV/AIDS mortality. Conclusion Rapid antiretroviral therapy can significantly reduce the risk of HIV/AIDS death. A sustainable model of rapid initiation of antiretroviral therapy should be further established to increase the proportion of rapid antiretroviral therapy for HIV/AIDS in Wuhan.
9.Quality control of Sagina japonica by HPLC fingerprint combined with quantitative analysis of multi-components by single-marker
Junhong LIU ; Xue LI ; Meiqin ZHANG ; Han HU ; Chunmei BAI ; Chunhua LIU ; Yongjun LI
China Pharmacy 2026;37(7):883-888
OBJECTIVE To establish the high-performance liquid chromatography (HPLC) fingerprint of Sagina japonica , and to establish a quantitative analysis of multi-components by single-marker (QAMS) method for simultaneous determination of six componen ts in S. japonica , aiming to provide references for the quality control of this medicinal herb. METHODS HPLC method was used to establish the fingerprints of 12 batches (No. S1-S12) of S . japonica according to Similarity Evaluation System of Chromatographic Fingerprint of Traditional Chinese Medicine . The similarity evaluation and identification of common peaks were conducted, followed by cluster analysis (CA) and principal component analysis (PCA) for 12 batches of samples. Using vicenin-2 as internal reference, the contents of p-hydroxy cinnamic acid, apigenin-6-C-arabinoside-8-C-glucoside, isoorientin, vitexin and 20-hydroxyecdysone were determined by QAMS method. The results were then compared with those obtained by the external standard method. RESULTS The similarities of HPLC fingerprints for 12 batches of S . japonica ranged from 0.828-0.998. A total of 17 common peaks were calibrated, and 6 common peaks were identified. Specifically, peak 5 was identified as vicenin-2, peak 7 as p-hydroxycinnamic acid, peak 10 as apigenin-6-C-arabinoside-8-C-glucoside, peak 11 as isoorientin, peak 13 as vitexin, and peak 15 as 20-hydroxyecdysone. The results of CA showed that S1-S5, S7 and S9-S11 were clustered into one category, S6 was clustered into one category, and S8 and S12 were clustered into one category. The results of PCA revealed that the accumulative contribution rate of the four main components was 89.430%. The content ranges measured by QAMS method for p-hydroxy cinnamic acid, apigenin-6-C-arabinoside-8-C-glucoside, isoorientin, vitexin and 20-hydroxyecdysone were 0.017 4-0.269 4, 0.568 8-4.240 3, 0.503 2-5.040 3, 0.024 0-0.132 0 and 2.551 3-4.881 1 mg/g, respectively. There was no significant difference in the contents of components measured between QAMS method and the external standard method ( P >0.05). CONCLUSIONS The established HPLC fingerprint and QAMS method can be used for quality evaluation and quality control of S . japonica.
10.Predictive value of serum DLL4 and CTRP5 levels for severity of diabetic retinopathy and visual disability
Yuan ZHAO ; Xiaodong LIU ; Yanling HAN
International Eye Science 2026;26(5):868-872
AIM:To investigating the predictive value of serum delta-like ligand 4(DLL4), complement C1q/tumor necrosis factor-related protein 5(CTRP5)levels for the severity of diabetic retinopathy(DR)and visual disability.METHODS:Patients with DR admitted to Tangshan Central Hospital between January 2022 and January 2024 were enrolled. Based on disease severity, patients were divided into a proliferative DR group and a non-proliferative DR group. After one year of follow-up, patients were further categorized into a vision disability group and a non-visual disability group based on visual impairment status. Enzyme-linked immunosorbent assay(ELISA)was used to detect serum levels of DLL4 and CTRP5. The data and serum levels of DLL4 and CTRP5 were compared among patients with different medical conditions and visual disabilities. Pearson method was used to explore the correlation between serum levels of DLL4, CTRP5 and glucose and lipid metabolism indicators. Multivariate Logistic regression was performed to explore the influencing factors of visual disability in DR Receiver operating characteristic(ROC)curve was performed to explore the value of serum levels of DLL4 and CTRP5 in predicting visual disability in DR.RESULTS: This study included 245 DR patients.Ninety-five patients were in the proliferative DR group(mean age 51.61±3.44 y, 51 men and 44 women), and 150 patients were in the non-proliferative DR group(mean age 51.22±3.11 y, 86 men and 64 women). The visually disability group consisted of 39 participants(mean age 51.64±3.87 y; 21 men and 18 women), while the non-visually disability group consisted of 206 participants(mean age 51.32±3.12 y; 116 men and 90 women). Patients in the proliferative DR group exhibited longer DR duration, higher levels of FPG, TG, TC, LDL-C, and serum DLL4 and CTRP5, and lower HDL-C levels compared to the non-proliferative DR group(all P<0.05). The serum levels of DLL4 were positively correlated with FPG(P<0.001), while the serum levels of CTRP5 were prominently positively correlated with FPG, TG, TC, LDL-C, and prominently negatively correlated with HDL-C(all P<0.001). The visual disability group had longer duration of DR and higher serum levels of DLL4 and CTRP5 than the non-visual disability group(all P<0.001). The duration of DR and serum levels of DLL4 and CTRP5 were influencing factors for visual disability in DR patients(all P<0.001). The joint detection of serum levels of DLL4 and CTRP5 had a higher value in predicting visual disability in DR patients than the single indicator prediction(ZDLL4-joint=3.018, PDLL4-joint=0.003; ZCTRP5-joint=2.784, PCTRP5-joint=0.005). CONCLUSION: Serum levels of DLL4 and CTRP5 are elevated in DR patients, and are closely related to the disease condition. The joint detection of serum levels of DLL4 and CTRP5 has high value in predicting visual disability in DR patients.


Result Analysis
Print
Save
E-mail