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.Study on the Mechanism of Panax Quinquefolium-Acorus Calamus Ameliorating Diabetic EncepHalopathy in Mice by Mediating Nrf2-Keap1 Signaling Pathway
Dezhi CUI ; You ZHOU ; Jianan LI ; Xu CHEN ; Linan HAN
Chinese Journal of Modern Applied Pharmacy 2024;41(9):1173-1182
OBJECTIVE
To observe the effects of Panax quinquefolium-Acorus calamus on learning and memory abilities in diabetes mellitus(DM) mice and investigate the mechanism of Panax quinquefolium-Acorus calamus in treating diabetic cognitive impairment(DCI) through network pharmacology and animal experiments.
METHODS
Diabetic mouse model was established by intraperitoneal injection of streptozotocin(80 mg·kg−1), followed by 8 weeks of oral administration and assessment of drug efficacy using the Morris water maze. The active ingredients and targets of Panax quinquefolium-Acorus calamus were collected using TCMSP, Swiss Target Prediction, and Gene Cards. The protein-protein interaction network of "Traditional Chinese Medicine-Ingredient-Disease targets" was constructed using the String platform and Cytoscape, visualized, and subjected to enrichment analysis using the Metascape database. The anti-DCI mechanism of Panax quinquefolium-Acorus calamus was examined through ELISA and Western blotting, while changes in hippocampal neurons of diabetic mice were observed using HE staining.
RESULTS
Panax quinquefolium-Acorus calamus reduced the escape latency of diabetic mice(P<0.05), without significant impact on swimming speed. Network pharmacology results indicated that the main components of Panax quinquefolium-Acorus calamus in treating DCI were ginsenoside Re, ginsenoside Rh2, and shanjin phenol, which regulated the Nrf2-Keap1 signaling pathway to treat DCI. Animal experiments demonstrated that Panax quinquefolium-Acorus calamus increased SOD activity(P<0.05), decreased MDA levels(P<0.01), enhanced the expression of HO-1, Keap1, Nrf2 in mouse brain(P<0.01), and alleviated the loosening of granule cell arrangement and nuclear condensation in the hippocampal CA1, CA3, and DG regions.
CONCLUSION
Using animal experiments combined with network pharmacology, this study preliminarily elucidates the potential targets and mechanisms of Panax quinquefolium-Acorus calamus in intervening DCI, and predictes the molecular basis for its intervention in DCI through molecular docking, providing insights for further in-depth research on Panax quinquefolium-Acorus calamus.
3. A comparative study on the use of a three-dimensional visualization operative planning system in Ultrasound-guided percutaneous microwave ablation for large hepatic hemangiomas
Yun CHEN ; Song ZHOU ; Xin LI ; Fangyi LIU ; Zhigang CHENG ; Jie YU ; Zhiyu HAN ; Linan DONG ; Ping LIANG ; Xiaoling YU
Chinese Journal of Hepatobiliary Surgery 2019;25(12):890-893
Objective:
To study the use of a three dimensional (3D) visualization operative planning system in Ultrasound-guided percutaneous microwave ablation (US-PMWA) for large hepatic hemangiomas (LHHs).
Methods:
A total of 50 patients with LHHs from January 2011 to August 2018 were included in Department of Interventional Ultrasound, the First Medical Center, Chinese PLA General Hospital, including 12 males and 38 females (age from 28.0~60.0, mean age was 43.0). Fifty patients with LHHs were divided into the 3D and 2D groups (25 cases in each group). The therapeutic efficacy was assessed by contrast-enhanced imagings on follow-up. Hepatic and renal functions were studied. The complete ablation, tumor volume shrinkage and complication rates were analyzed.
Results:
The levels of alanine aminotransferase, aspartate aminotransferase, alkaline phosphatase, and creatinine after ablation in the 3D group were significantly lower than the 2D group [(126.7±56.4)U/L vs. (204.7±76.5)U/L; (141.0±60.8)U/L vs. (206.6±77.4)U/L; (57.3±17.6)U/L vs. (86.2±46.1)U/L; (66.6±16.6)mmol/L vs. (86.8±42.8)mmol/L,
4.Efficacy and safety of drospirenone-ethinylestradiol on contraception in healthy Chinese women: a multicenter randomized controlled trial
Guangsheng FAN ; Meilu BIAN ; Linan CHENG ; Xiaoming CAO ; Zirong HUANG ; Ziyan HAN ; Xiaoping JING ; Jian LI ; Shuying WU ; Chengliang XIONG ; Zhengai XIONG ; Tianfu YUE
Chinese Journal of Obstetrics and Gynecology 2009;44(1):38-44
Objective To evaluate the contraception efficacy, mode of bleeding, side effects and other positive effects of drospirenone-ethinylestradiol (Yasmin) in healthy Chinese women. Methods This was a multicenter, randomized, control study of 768 healthy Chinese women who consulted about contraception. The subjects were randomized into Yasmin group (30 μg ethinylestradiol plus 3 mg drospirenone, 573 cases) or desogestrel group (30 μg ethinylestradiol plus 150 μg desogestrel, 195 cases) with the ratio of 3: 1. Each individual was treated for 13 cycles. Further visits were required at cycle 4, cycle 7, cycle 10 and cycle 13 of treatment. Weight, height, body mass index were evaluated at each visit. The menstrual distress questionnaire (MDQ) was given to the women at baseline, visit 3 (cycle 7) and visit 5 (after cycle 13). Results The values of basal features were similar between two groups (P> 0.05). The Pearl index (method failure) of Yasmin was 0. 208/hundred women year which was lower than that of desogestrel (0. 601/hundred women year). The mode of bleeding was similar between two groups after trial without showing any significant difference. According to MDQ subscale, the improvement of water retention and increasing appetite during inter-menstrual period and water retention and general well-being during menstrual period in the Yasmin group ( -0. 297, -0. 057, 0. 033, 0. 150 respectively) was more obvious than that in the desogestrel group ( - 0. 108, 0. 023, 0. 231, - 0. 023 respectively) with a significant difference (P < 0. 05 ). Some other values which improved in beth two groups, especially the improvement of breast tenderness and pain and skin abnormality in Yasmin group (18.0%, 89/494; 12. 6%, 62/494) was more distinct than that in desogestrel group (11.3%, 19/168; 5.4%, 9/168). The mean weight increased in desogestrel group (0. 57 kg) while it decreased in Yasmin group ( -0. 28 kg) with a significant difference (P < 0. 01 ). Conclusions Both Yasmin and desogestrel have good efficacy on contraception and similar modes of menstrual bleeding. Yasmin is better than desogestrel in terms of weight control and premenstrual syndrome of oral contraceptive.
Result Analysis
Print
Save
E-mail