ObjectiveTo develop a predictive model and application for spontaneous passage of common bile duct stones using automated machine learning algorithms given the complexity of treatment decision-making for patients with common bile duct stones， and to reduce unnecessary endoscopic retrograde cholangiopancreatography （ERCP） procedures. MethodsA retrospective analysis was performed for the data of 835 patients who were scheduled for ERCP after a confirmed diagnosis of common bile duct stones based on imaging techniques in Changshu First People’s Hospital （dataset 1） and Changshu Traditional Chinese Medicine Hospital （dataset 2）. The dataset 1 was used for the training and internal validation of the machine learning model and the development of an application， and the dataset 2 was used for external testing. A total of 22 potential predictive variables were included for the establishment and internal validation of the LASSO regression model and various automated machine learning models. The area under the receiver operating characteristic curve （AUC）， sensitivity， specificity， and accuracy were used to assess the performance of models and identify the best model. Feature importance plots， force plots， and SHAP plots were used to interpret the model. The Python Dash library and the best model were used to develop a web application， and external testing was conducted using the dataset 2. The Kolmogorov-Smirnov test was used to examine whether the data were normally distributed， and the Mann-Whitney U test was used for comparison between two groups， while the chi-square test or the Fisher’s exact test was used for comparison of categorical data between groups. ResultsAmong the 835 patients included in the study， 152 （18.20%） experienced spontaneous stone passage. The LASSO model achieved an AUC of 0.875 in the training set （n=588） and 0.864 in the validation set （n=171）， and the top five predictive factors in terms of importance were solitary common bile duct stones， non-dilated common bile duct， diameter of common bile duct stones， a reduction in serum alkaline phosphatase （ALP）， and a reduction in gamma-glutamyl transpeptidase （GGT）. A total of 55 models were established using automated machine learning， among which the gradient boosting machine （GBM） model had the best performance， with an AUC of 0.891 （95% confidence interval： 0.859‍ ‍—‍ ‍0.927）， outperforming the extreme randomized tree mode， the deep learning model， the generalized linear model， and the distributed random forest model. The GBM model had an accuracy of 0.855， a sensitivity of 0.846， and a specificity of 0.857 in the test set （n=76）. The variable importance analysis showed that five factors had important influence on the prediction of spontaneous stone passage， i.e.， were solitary common bile duct stones， non-dilated common bile duct， a stone diameter of <8 mm， a reduction in serum ALP， and a reduction in GGT. The SHAP analysis of the GBM model showed a significant increase in the probability of spontaneous stone passage in patients with solitary common bile duct stones， non-dilated common bile duct， a stone diameter of <8 mm， and a reduction in serum ALP or GGT. ConclusionThe GBM model and application developed using automated machine learning algorithms exhibit excellent predictive performance and user-friendliness in predicting spontaneous stone passage in patients with common bile duct stones. This application can help avoid unnecessary ERCP procedures， thereby reducing surgical risks and healthcare costs.

ObjectiveTo investigate the independent risk factors for poor prognosis in patients with acute pancreatitis （AP） by analyzing inflammatory factors， lung ultrasound （LUS） scores， and CT scores， to establish a nomogram prediction model， and to provide a basis for early clinical intervention. MethodsA total of 409 patients with AP who were admitted to Changshu Hospital Affiliated to Soochow University from January 2021 to October 2023 were enrolled as subjects， and they were divided into modeling group with 288 patients and validation group with 121 patients using the simple random sampling method at a ratio of 7∶3. According to the prognosis， each group was further divided into poor prognosis group and good prognosis group. The levels of C-reactive protein （CRP）， procalcitonin （PCT）， interleukin-6 （IL-6）， interleukin-10 （IL-10）， and tumor necrosis factor-α （TNF-α） were measured for both groups within 72 hours after admission， and LUS scores， modified CT severity index （MCTSI）， and extrapancreatic inflammation on computed tomography （EPIC） scores were assessed within 48‍ ‍—‍ ‍72 hours after admission. The independent-samples t test was used for comparison of normally distributed continuous data between groups， and the Mann-Whitney U rank sum test was used for comparison of non-normally distributed continuous data between groups； the chi-square test was used for comparison of categorical data between groups. A LASSO regression analysis was used to screen for the variables that were included in the multivariate logistic regression model to identify the independent risk factors for the poor prognosis of AP， and then a nomogram prediction model was established. The receiver operating characteristic （ROC） curve and the calibration curve were used to assess the discriminatory ability and goodness of fit of the nomogram model， and a decision curve analysis was used to assess the clinical applicability of the model. ResultsAmong the 288 patients with AP in the modeling group， there were 33 （11.46%） in the poor prognosis group and 255 （88.54%） in the good prognosis group； among the 121 patients with AP in the validation group， there were 13 （10.74%） in the poor prognosis group and 108 （89.26%） in the good prognosis group. Compared with the good prognosis group， the poor prognosis group had significantly higher levels of CRP （Z=3.607， P<0.05）， IL-6 （Z=4.189， P<0.05）， and TNF-α （t=2.584， P<0.05）， and significantly higher scores of LUS （t=8.075， P<0.05）， MCTSI （t=5.929， P<0.05）， and EPIC （t=8.626， P<0.05）. The multivariate logistic regression analysis showed that CRP （odds ratio ［OR］=3.592， 95% confidence interval ［CI］： 1.272‍ ‍—‍ ‍10.138， P<0.05）， IL-6 （OR=4.225， 95%CI： 1.468‍ ‍—‍ ‍12.156， P<0.05）， TNF-α （OR=3.540， 95%CI： 1.205‍ ‍—‍ ‍10.401， P<0.05）， LUS （OR=7.094， 95%CI： 2.398‍ ‍—‍ ‍20.986， P<0.05）， MCTSI （OR=7.612， 95%CI： 2.832‍ ‍—‍ ‍20.462， P<0.05）， and EPIC （OR=11.915， 95%CI： 4.007‍ ‍—‍ ‍35.432， P<0.05） were independent risk factor for poor prognosis in patients with AP. A nomogram prediction model was established based on the above 6 indicators， which had an area under the ROC curve of 0.924 （95%CI： 0.883‍ ‍—‍ ‍0.964）， and the Youden index for the optimal cut-off value was 0.670， with a sensitivity of 0.909 and a specificity of 0.761. The calibration curve showed good consistency between the predicted and observed results in both the modeling group and the validation group. The decision curve analysis showed that the predictive model had certain clinical effectiveness. ConclusionThe nomogram model for predicting the risk of poor prognosis in AP patients based on CRP， IL-6， TNF-α， LUS score， MCTSI score， and EPIC score has relatively good predictive performance and can provide important strategic guidance for developing early intensified treatment regimens for AP patients in clinical practice.

Objective·To compare the performance of various deep learning methods in the segmentation and classification of colorectal polyp endoscopic images,and identify the most effective approach.Methods·Four colorectal polyp datasets were collected from three hospitals,encompassing 1 534 static images and 15 videos.All samples were pathologically validated and categorized into two types:serrated lesions and adenomatous polyps.Polygonal annotations were performed by using the LabelMe tool,and the annotated results were converted into integer mask formats.These data were utilized to train various architectures of deep neural networks,including convolutional neural network(CNN),Transformers,and their fusion,aiming to develop an effective semantic segmentation model.Multiple performance indicators for automatic diagnosis of colon polyps by different architecture models were compared,including mIoU,aAcc,mAcc,mDice,mFscore,mPrecision and mRecall.Results·Four different architectures of semantic segmentation models were developed,including two deep CNN architectures(Fast-SCNN and DeepLabV3plus),one Transformer architecture(Segformer),and one hybrid architecture(KNet).In a comprehensive performance evaluation of 291 test images,KNet achieved the highest mIoU of 84.59％,significantly surpassing Fast-SCNN(75.32％),DeepLabV3plus(78.63％),and Segformer(80.17％).Across the categories of"background","serrated lesions"and"adenomatous polyps",KNet's intersection over union(IoU)were 98.91％,74.12％,and 80.73％,respectively,all exceeding other models.Additionally,KNet performed excellently in key performance metrics,with aAcc,mAcc,mDice,mFscore,and mRecall reaching 98.59％,91.24％,91.31％,91.31％,and 91.24％,respectively,all superior to other models.Although its mPrecision of 91.46％was not the most outstanding,KNet's overall performance remained leading.In inference testing on 80 external test images,KNet maintained an mIoU of 81.53％,demonstrating strong generalization capabilities.Conclusion·The semantic segmentation model of colorectal polyp endoscopic images constructed by deep neural network based on KNet hybrid architecture,exhibits superior predictive performance,demonstrating its potential as an efficient tool for detecting colorectal polyps.

Objective To develop deep learning models for colonoscopy quality control using various deep learning architectures,and to delve into the decision-making mechanisms.Methods The colonoscopy images were selected from two datasets separately constructed by the HyperKvasir and Changshu Hospital Affiliated to Soochow University,encompassing intestines of varying degrees of cleanliness,polyps,and cecums.After image preprocessing and enhancement,transfer learning was carried out using the pre-trained models based on convolutional neural network(CNN)and Transformer.The model training adopted cross-entropy loss functions and Adam optimizer,and simultaneously implemented learning rate scheduling.To enhance model transparency,a thorough interpretability analysis was conducted using Grad-CAM,Guided Grad-CAM,and SHAP.The final model was converted to ONNX format and deployed on various equipment terminals to achieve real-time colonoscopy quality control.Results In a dataset of 3 831 colonoscopy images,EfficientNet model outperformed the other models on the test set,achieving an accuracy of 0.992 which was higher than those of the other models based on CNN(DenseNet121,ResNet50,VGG19)and Transformer(ViT,Swin,CvT),with a precision,recall rate,and F1 score of 0.991,0.989,and 0.990.On an external test set of 358 images,EfficientNet model had an average AUC,precision,and recall rate of 0.996,0.948,and 0.952,respectively.Although EfficientNet model is high-performing,some misjudgments still occurred.Interpretability analysis highlighted key image areas affecting decision-making.In addition,EfficientNet model was successfully converted to ONNX format and deployed on multiple platforms and devices,and it ensured real-time colonoscopy quality control with an inference speed of over 60 frames per second.Conclusion Among the 7 models developed for colonoscopy quality control based on CNN and Transformer,EfficientNet demonstrated exemplary performance across all categories and is deployed for real-time predictions on multiple terminals,aiming to provide patients with better medical care.

Objective:To evaluate the impacts of resistance training(RT)and aerobic training(AT)for 24 months on the risk of type 2 diabetes in patients with pre-diabetes.Methods:Two hundred forty-eight pre-diabetic patients were enrolled in this multi-center randomized controlled trial. All patients were randomly divided into 3 groups: RT( n=82), AT( n=83), and control( n=83)groups. The participants in RT and AT groups undertook moderate RT or AT 3 times a week(150 minutes/week)under supervision in 3 research centers for 24 months. Elastic bands were used in each session of RT, with intensity of 60% 1RM(maximum weight that muscle can lift at once). Patients in AT group performed aerobic dance at 60%-70% of maximum heart rate. Assessments for each subject were made at baseline and by the end of 6, 12 and 24 months. Primary outcomes were changes in the risk of type 2 diabetes. Secondary outcomes included changes in blood glucose, blood lipids, and blood pressure. Results:There were 217, 206, and 173 subjects who completed the follow-up of 6, 12, and 24 months, respectively. The mean ages of RT, AT, and control groups at baseline were(59.91±5.92), (60.93±5.71), and(60.73±5.83)years. Compared to control group, both RT and AT groups revealed a significant reduction in HbA 1C( P<0.05), and a significant increase in homeostasis model assessment for β-cell function index(HOMA2-β, P<0.01)by the end of 12 and 24 months. Adjusted for age, gender, statin use, lipid profile, blood pressure, and body mass index, COX regression analysis showed that RT and AT reduced the risk of type 2 diabetes by 55.6%( P=0.012)and 59.8%( P=0.010). Conclusions:This study demonstrates that 24-month moderate RT and AT have comparable effects on reducing insulin resistance, improving β-cell function, blood glucose and lipid, and reducing the risk of type 2 diabetes.

Objective:To investigate the clinical effect of free deep femoral artery third perforating flap repaired soft tissue loss after Pilon fracture surgery in I stage.Methods:Fifteen patients were treated from April, 2013 to January, 2020. Miller AO classification: 8 cases 43-C1, 4 cases 43-C2 and 3 cases 43-C3. All cases were accompanied with severe soft tissue contusion and skin necrosis. After fracture reduction, soft tissue defects, internal fixation exposure and tendon exposure around the wound. Free deep femoral artery third perforating flap (3.5 cm ×15.5 cm to 5.5 cm×12.5 cm) for the repair of soft tissue defects around ankle in the I stage, the blood vessels of the flap were end-to-side anastomosed with vessels of the posterior tibial or anterior tibial. Regular follow-up after surgery.Results:One case of venous crisis occurred, other 14 cases survived, were followed-up from 5 to 18 months, the ankle joint function was good, did not affect the foot shoes, with excellent color and texture, the flap restored protective sensation, and leaving only linear scar, no muscle adhesion.Conclusion:Free deep femoral artery third perforating flap repaired soft tissue loss of surgical incision after fracture operated than significantly reduce the postoperative fracture infection and protect the blood supply around the fracture. It is an effective method of repair.

Objective:To investigate the effect of a 2-year resistance and aerobic training on reducing the risk of cardiovascular disease in patients with prediabetes.Methods:A total of 248 patients with prediabetes were enrolled from Chinese and Western Medicine Hospital Affiliated to Nanjing University of Chinese Medicine from January to April 2014, and Danyang People′s Hospital and The First Affiliated Hospital of Guangxi Medical University from May to December 2014.Based on random number table method, the patients were divided into 3 groups: the resistance training group (RT group, 82 cases), the aerobic training group (AT group, 83 cases) and control group (83 cases). Participants in the RT group and the AT group underwent a total of 24 months of exercise training. Changes in indicators (blood glucose,blood lipid, etc.) at baseline and the end of 12 and 24 months among the groups were compared.Results:After intervention, glycosylated hemoglobin (HbA1c), low density lipoprotein cholesterol (LDL-C), blood pressure and homeostasis model 2 insulin resistance index (HOMA2-IR) in the RT and AT groups tended to decrease, and the steady state model 2 β cell function index (HOMA2-β) tended to increase. At the end of 24 months, HbA1c [5.80 (5.43, 6.20) %, 5.70 (5.50, 6.00)% vs. 6.20 (5.70, 6.60) %, all P≤ 0.01], LDL-C [3.07 (2.69, 3.58) mmol/L, 2.97 (2.62, 3.95) mmol/L vs. 3.21(2.54, 3.78) mmol/L, all P<0.05] and HOMA2-IR [0.96 (0.82, 1.47), 1.20 (0.99, 1.43) vs. 1.34 (1.09, 1.51), all P<0.05] were significantly decreased in the RT and AT groups than in the control group. In addition, HOMA2-β [84.50 (60.55, 107.33), 93.00 (78.60, 119.75) vs. 53.40 (37.70, 80.40), all P = 0.001] was significantly increased in the AT and RT groups compared with that in the control group. There were no significant differences in triglyceride (TG) and high-density lipoproteincholesterol (HDL-C) levels between the training groups and the control group (all P>0.05). After adjusting for age, sex and blood pressure, the cardiovascular risk of prediabetes was significantly reduced in RT ( P =0.017) and AT groups ( P =0.018). The Cox regression analyses showed that both the resistance training (HR=0.419, 95 %CI =0.415-0.942, P=0.037) and the aerobic training ( HR=0.310, 95 %CI=0.447-0.866, P=0.026) were protective factors for cardiovascular disease in prediabetic patients after adjustment of age, sex, statins, body mass index and waist-to-hip ratio, which reduced the risks of cardiovascular disease in prediabetic patients by 58.1% and 69.0%, respectively. Conclusions:Two years of aerobic and resistance training interventions have obvious advantages on glycemic and insulin resistance control in prediabetes patients. The resistance training can reduce the risk of cardiovascular disease, and it is, thus, recommended for prediabetic patients without obvious exercise contraindications.

Objective:The research was aimed to investigate the association between serum total homocysteine (tHcy) and subacute combined degeneration of the spinal cord (SCD).Methods:A retrospective survey of 106 newly diagnosed patients with SCD were enrolled in this research who were treated in the department of neurology of Xijing Hospital from January 2008 to February 2019, meanwhile, 121 patients with spinal cord lesion (not SCD) and 104 neurology mild outpatients were selected as controls. Serum tHcy level was determined by using the chemiluminescent immunoassay assay. A multivariate logistic regression model was used to analyze the risk factors for SCD. The area under the curve (AUC) of the receiver operating characteristic (ROC) curve, sensitivity, specificity and Youden index were used to evaluate the diagnostic efficacy of tHcy. Spearman correlation analysis was used to observe the correlation between tHcy and SCD severity. The SCD patients were categorized into normal or mild tHcy group, moderate tHcy group, and severe tHcy group based on tHcy levels. Clinical symptoms, nerve conduction velocity, magnetic resonance imaging (MRI) findings from the patients were studied.Results:The serum tHcy levels in SCD patients were 64.3(26.5, 98.8) μmol/L, while in patients with spinal cord lesion (not SCD) group were 13.7(10.8, 19.2) μmol/L, neurology mild outpatients were 10.6(8.2, 13.0) μmol/L, which was higher in SCD group ( H=112.020, P<0.001), ( H=165.525, P<0.001).The multivariate logistic regression model showed tHcy is the impact factor of SCD ( OR=1.107, 95% CI:1.077-1.139, P<0.001). At ROC analysis, tHcy showed diagnostic value with an optimal cut-off value of 24.9 μmol/L (AUC 0.913, 95% CI: 0.875-0.951, sensitivity 79.2%, specificity 91.6%). Spearman correlation analysis showed that tHcy was positively correlated with functional disability rating scale ( r=0.254, P=0.009). Conclusions:Serum tHcy is the risk factor for SCD and related to its disability. Focus on the increased level of tHcy plays a positive role in the diagnosis of SCD.

Objective:To evaluate the prognosis and determine the failure patterns after radiotherapy for low-risk early-stage patients with extranodal NK/T-cell lymphoma, nasal-type (ENKTCL).Methods:A total of 557 patients from 2000—2015 with low-risk early-stage ENKTCL who received radiotherapy (RT) with or without chemotherapy (CT) from China Lymphoma Collaborative Group were retrospectively reviewed. Among them, 427 patients received combined modality therapy, whereas 130 patients received RT alone. Survivals were calculated by Kaplan-Meier method and compared with Log-rank test. Overall survival (OS) was compared with age and sex-matched general Chinese population using expected survival and standardized mortality ratio (SMR). Cox stepwise regression model was used for multivariate analysis.Results:The 5-year OS and progression-free survival (PFS) were 87.2% and 77.2%. The SMR was 3.59 ( P<0.001) at 1 year after treatment, whereas it was 1.50 at 4 years after treatment, without significant difference between ENKTCL group and country-matched general population ( P=0.146). Compared with RT alone, CMT did not result in significantly superior 5-year OS (87.0% vs 87.4%, P=0.961) or PFS (76.1% vs 80.7%, P=0.129). Local failure (11.5%, 64/557) and distant failure (10.8%, 60/557) were the main failure modes, while regional failure was rare (2.9%, 16/557). The 5-year locoregional control rate (LRC) was 87.2% for the whole group, with 89.5% for ≥50 Gy versus 73.7% for <50 Gy ( P<0.001). Radiotherapy dose was an independent factor affecting LRC( P<0.05). Conclusions:Radiotherapy achieves a favorable prognosis in patients with low-risk early-stage ENKTCL. The incidence of either locoregional or distant failure is low. Radiation dose still is an important prognostic factor for LRC.

Objective:To evaluate the prognosis and determine the failure patterns after radiotherapy for low-risk early-stage patients with extranodal NK/T-cell lymphoma, nasal-type (ENKTCL).Methods:A total of 557 patients from 2000—2015 with low-risk early-stage ENKTCL who received radiotherapy (RT) with or without chemotherapy (CT) from China Lymphoma Collaborative Group were retrospectively reviewed. Among them, 427 patients received combined modality therapy, whereas 130 patients received RT alone. Survivals were calculated by Kaplan-Meier method and compared with Log-rank test. Overall survival (OS) was compared with age and sex-matched general Chinese population using expected survival and standardized mortality ratio (SMR). Cox stepwise regression model was used for multivariate analysis.Results:The 5-year OS and progression-free survival (PFS) were 87.2% and 77.2%. The SMR was 3.59 ( P<0.001) at 1 year after treatment, whereas it was 1.50 at 4 years after treatment, without significant difference between ENKTCL group and country-matched general population ( P=0.146). Compared with RT alone, CMT did not result in significantly superior 5-year OS (87.0% vs 87.4%, P=0.961) or PFS (76.1% vs 80.7%, P=0.129). Local failure (11.5%, 64/557) and distant failure (10.8%, 60/557) were the main failure modes, while regional failure was rare (2.9%, 16/557). The 5-year locoregional control rate (LRC) was 87.2% for the whole group, with 89.5% for ≥50 Gy versus 73.7% for <50 Gy ( P<0.001). Radiotherapy dose was an independent factor affecting LRC( P<0.05). Conclusions:Radiotherapy achieves a favorable prognosis in patients with low-risk early-stage ENKTCL. The incidence of either locoregional or distant failure is low. Radiation dose still is an important prognostic factor for LRC.