BACKGROUND: Over 20 million colonoscopies are performed in China annually. An automatic clinical decision support system (CDSS) with accurate semantic recognition of colonoscopy reports and guideline-based is helpful to relieve the increasing medical burden and standardize the healthcare. In this study, the CDSS was built under a hierarchical-label interpretable classification framework, trained by a state-of-the-art transformer-based model, and validated in a multi-center style. METHODS: We conducted stratified sampling on a previously established dataset containing 302,965 electronic colonoscopy reports with pathology, identified 2041 patients' records representative of overall features, and randomly divided into the training and testing sets (7:3). A total of five main labels and 22 sublabels were applied to annotate each record on a network platform, and the data were trained respectively by three pre-training models on Chinese corpus website, including bidirectional encoder representations from transformers (BERT)-base-Chinese (BC), the BERT-wwm-ext-Chinese (BWEC), and ernie-3.0-base-zh (E3BZ). The performance of trained models was subsequently compared with a randomly initialized model, and the preferred model was selected. Model fine-tuning was applied to further enhance the capacity. The system was validated in five other hospitals with 3177 consecutive colonoscopy cases. RESULTS: The E3BZ pre-trained model exhibited the best performance, with a 90.18% accuracy and a 69.14% Macro-F1 score overall. The model achieved 100% accuracy in identifying cancer cases and 99.16% for normal cases. In external validation, the model exhibited favorable consistency and good performance among five hospitals. CONCLUSIONS The novel CDSS possesses high-level semantic recognition of colonoscopy reports, provides appropriate recommendations, and holds the potential to be a powerful tool for physicians and patients. The hierarchical multi-label strategy and pre-training method should be amendable to manage more medical text in the future.
Humans ; Colonoscopy/methods* ; Deep Learning ; Decision Support Systems, Clinical ; Female ; Male

In recent years， clinical studies on targeted therapy and immunotherapy for advanced hepatocellular carcinoma used alone or in combination have provided abundant evidence on efficacy and safety for the selection of first-line therapies. However， no consensus has been reached on the selection of second-line therapies in various clinical guidelines for hepatocellular carcinoma， which is caused by the fact that existing evidence is limited to the options after failure of sorafenib and that there is still a lack of high-level evidence for new first-line therapies such as second-line therapies after resistance to targeted therapy and immunotherapy for hepatocellular carcinoma. This article reviews the results of current clinical trials and summarizes the studies on second-line therapies for hepatocellular carcinoma after resistance to first-line targeted therapy and immunotherapy for hepatocellular carcinoma based on the different mechanisms of action of drugs， as well as the research advances in recent years. For hepatocellular carcinoma patients with resistance to first-line targeted therapy and immunotherapy， targeted combination therapy and dual-immune therapy are expected to improve treatment outcome and survival， and more prospective clinical studies are needed in the future to provide effective and safe treatment regimens for hepatocellular carcinoma patients with resistance to targeted therapy and immunotherapy.

Artificial chord is a mature mitral valve repair technique, especially in adult mitral valve repair. It is still challenging to repair mitral valve in children with artificial chords because the quality of mitral valve is soft and immature. There are some differences in the methods of suture, the choice of suture size and the number of artificial chords. Although the artificial chords could not grow naturally, we found through the long-term research that most children did not have mitral valve restriction or even chords rupture due to itself can compensate through the growth of the flap and papillary muscle. This article summarizes the recent research progress on the treatment of mitral valve insufficiency in children with artificial chords, providing reference for clinical treatment.

Purpose To explore the value of deep learning feature prediction based on the ResNet50 deep residual network model for predicting early hematoma expansion in spontaneous intracerebral hemorrhage without traditional imaging manifestations. Materials and Methods A retrospective study was performed on 235 patients with spontaneous intracerebral hemorrhage in Jiangdu People's Hospital Affiliated to Yangzhou University from January 2019 and December 2022. These patients had undergone their initial plain cranial CT scan within 6 hours of symptom onset and a subsequent follow-up scan within 24 hours of admission. They were randomly assigned to a training set consisting of 188 cases and a test set of 47 cases,using an 8︰2 ratio. The region of interest (ROI) of hematoma was traced layer by layer on the first plain head CT,and image genomics features were extracted. The maximum two-dimensional cross-sectional ROI of the hematoma 3D-ROI,as well as ROI images at 1 mm and 2 mm above and below the maximum two-dimensional cross-sectional ROI,were then cut and input into the pre-trained ResNet50 model for feature extraction. The image genomics features were then fused with the extracted deep learning features using a least absolute shrinkage and selection operator regression model. A support vector machine classifier was used to construct a prediction model,which was evaluated using receiver operating characteristic curves and decision curve analysis. Results In the training set,the area under curve (AUC) of the deep learning feature model was 0.972,which was higher than that of the image genomics feature model (0.951) and the fused feature model (0.968),but this difference was not statistically significant (P>0.05). In the testing set,the AUCs of the deep learning feature model and the fused feature model were 0.867 and 0.895,respectively,which were significantly higher than that of the image genomics feature model (0.833),with statistically significant differences (Z=-1.794,-2.191,both P<0.05). The AUC of the fused feature model showed an improvement compared to the deep learning feature model,but the difference was not statistically significant (P>0.05). In the test set,decision curve analysis revealed that the fused feature model yielded greater benefits compared to both the deep learning feature model and the radiomic feature model. Conclusion The deep learning feature model based on ResNet50 deep residual network shows better performance in predicting early hematoma expansion than the image genomics feature model,and the fused feature model has a beneficial effect on predicting hematoma expansion. This deep learning approach provides a prediction tool with supervisory capability for clinical decision-making.

Purpose To explore the value of deep learning feature prediction based on the ResNet50 deep residual network model for predicting early hematoma expansion in spontaneous intracerebral hemorrhage without traditional imaging manifestations. Materials and Methods A retrospective study was performed on 235 patients with spontaneous intracerebral hemorrhage in Jiangdu People's Hospital Affiliated to Yangzhou University from January 2019 and December 2022. These patients had undergone their initial plain cranial CT scan within 6 hours of symptom onset and a subsequent follow-up scan within 24 hours of admission. They were randomly assigned to a training set consisting of 188 cases and a test set of 47 cases,using an 8︰2 ratio. The region of interest (ROI) of hematoma was traced layer by layer on the first plain head CT,and image genomics features were extracted. The maximum two-dimensional cross-sectional ROI of the hematoma 3D-ROI,as well as ROI images at 1 mm and 2 mm above and below the maximum two-dimensional cross-sectional ROI,were then cut and input into the pre-trained ResNet50 model for feature extraction. The image genomics features were then fused with the extracted deep learning features using a least absolute shrinkage and selection operator regression model. A support vector machine classifier was used to construct a prediction model,which was evaluated using receiver operating characteristic curves and decision curve analysis. Results In the training set,the area under curve (AUC) of the deep learning feature model was 0.972,which was higher than that of the image genomics feature model (0.951) and the fused feature model (0.968),but this difference was not statistically significant (P>0.05). In the testing set,the AUCs of the deep learning feature model and the fused feature model were 0.867 and 0.895,respectively,which were significantly higher than that of the image genomics feature model (0.833),with statistically significant differences (Z=-1.794,-2.191,both P<0.05). The AUC of the fused feature model showed an improvement compared to the deep learning feature model,but the difference was not statistically significant (P>0.05). In the test set,decision curve analysis revealed that the fused feature model yielded greater benefits compared to both the deep learning feature model and the radiomic feature model. Conclusion The deep learning feature model based on ResNet50 deep residual network shows better performance in predicting early hematoma expansion than the image genomics feature model,and the fused feature model has a beneficial effect on predicting hematoma expansion. This deep learning approach provides a prediction tool with supervisory capability for clinical decision-making.

Objective:To investigate the prognosis and risk factors for children diagnosed with all types of subaortic stenosis(SAS) who developed recurrent left ventricular outflow tract obstruction after surgical treatment.Methods:The study retrospectively included patients aged 0-18 years old who underwent open heart SAS surgery at Fuwai Hospital from 2016-2019. Children with hypertrophic obstructive cardiomyopathy were excluded. Detailed operative notes, medical records and ultrasound information, and follow-ups were extracted. Recurrent SAS was defined as left ventricular outflow tract gradient 30 mmHg(1 mmHg=0.133 kPa) 1 month after SAS surgical treatment.Results:A total of 137 children were included in this study. The medium age of children at the time of SAS surgery was 4.6 years old(3 months-17.8 years old). After a median follow-up of 4.36 years(3.2-5.7 years), a total of 30 patients developed recurrent LVOTO, with a recurrence rate of 21.9%, and 7(5.1%) underwent a second surgery. Compared to the non-recurrent group, children in the recurrent group were younger at the time of surgery( P=0.0443), had a smaller body surface area( P=0.0485), and a longer length of stay( P=0.0380). In Cox analysis, when only considering preoperative variables, the independent risk factor for LVOTO recurrence were a peak left ventricular outflow tract gradient higher than 50 mmHg( HR=5.25, P=0.001), a BSA less than 0.9( HR=2.5, P=0.023), and a length of SAS 5 mm( HR=2.29, P=0.050). When both preoperative and intraoperative variables were considered, preoperative peak left ventricular outflow tract gradient 50 mmHg( HR=4.91, P=0.002) and peeling from the aortic valve( HR=3.23, P=0.010) were independent risk factors for postoperative recurrence. Conclusion:Recurrent LVOTO after SAS surgical repair is common, and regular postoperative follow-up is crucial to evaluate whether a secondary intervention is required. Regular postoperative follow-up is needed for children at high risk.

Objective     To summarize the characteristics of children diagnosed with secondary subaortic stenosis after the surgical closure for ventricular septal defect and explore its potential mechanism. Methods     We retrospectively collected patients aged from 0 to 18 years, who underwent ventricular septal defect closure and developed secondary subaortic stenosis, and subsequently received surgical repair from 2008 to 2019 in Fuwai Hospital. Their surgical details, morphological features of the subaortic stenosis, and the follow-up information were analyzed. Results     Six patients, including 2 females and 4 males, underwent the primary ventricular septal defect closure at the median age of 9 months (ranging from 1 month to 3 years). After the first surgery, patients were diagnosed with secondary subaortic stenosis after 2.9 years (ranging from 1 to 137 months). Among them, 2 patients underwent the second surgery immediately after diagnosis, and the other 4 patients waited 1.2 years (ranging from 6 to 45 months) for the second surgery. The most common type of the secondary subaortic stenosis after ventricular septal defect closure was discrete membrane, which located underneath the aortic valve and circles as a ring. In some patients, subaortic membrane grew along with the ventricular septal defect closure patch. During the median follow-up of 8.1 years (ranging from 7.3 to 8.9 years) after the sencond surgery, all patients recovered well without any recurrence of left ventricular outflow tract obstruction. Conclusion     Regular and persistent follow-up after ventricular septal defect closure combining with or without other cardiac malformation is the best way to diagnose left ventricular outflow tract obstruction in an early stage and stop the progression of aortic valve regurgitation.

Objective To observe the value of subregional non-contrast CT(NCCT)radiomics features based on habitat imaging technology for predicting hematoma expansion(HE)in patients with spontaneous intracranial hemorrhage(sICH).Methods Data of 228 sICH patients with negative conventional imaging signs were retrospectively analyzed and divided into HE group(n=99)or non HE(NHE)group(n=129)based on the occurrence of HE nor not.also divided into training set(n=182)or test set(n=46)at a ratio of 8:2.Clinical data,NCCT data and laboratory examination results were compared between groups.Logistic regressive analysis was performed to screen the impact factors of HE.ROI of whole hematoma(ROIwhole)was sketched and clustered into 3 sub-regions(ROIsub1,ROIsub2 and ROIsub3,the latter located in the critical area between hematoma and brain tissue)with habitat imaging technology,and radiomics features of ROI were extracted and screened.Then 4 prediction models were constructed based on the above 4 ROI,and the efficacy of each model for predicting HE was analyzed.Results The fasting blood glucose in HE group was higher than that in NHE group(t=2.047,P=0.041),which was not independent impact factor for predicting HE in sICH patients(P=0.070)according to logistic regression analysis.The area under the curve of ROIsub3 radiomics model for predicting sICH HE in training and test set was 0.945 and 0.863,respectively,not significantly different with that of ROIwhole(0.921,0.813),ROIsub1(0.925,0.807)nor ROIsub2(0.909,0.720)(all P＞0.05).Decision curve analysis showed that ROIsub3 radiomics model could bring greater benefits than the other 3 models.Conclusion NCCT radiomics features of the critical area between hematoma and brain tissue based on habitat imaging technology had high value for predicting HE in sICH patients.

Objective:To explore the protective effect and safety of RotaTeq vaccine on children with rotavirus gastroenteritis (RVGE) in high mortality areas in the world and guide the correct use of RotaTeq vaccine.Methods:The literature on RotaTeq vaccine in high mortality areas in the world published from February 2006 to December 2021 was searched, screened and sorted out according to the exclusion and inclusion criteria, and the data were analyzed by RevMan 5.3, Stata 14.0 and SPSS 26.0 softwares.Results:A total of 5 reports were enrolled, including 63 974 subjects, including 32 092 subjects in the vaccine group and 31 882 subjects in the placebo group. In high mortality areas, the protection rates of RotaTeq vaccine against RVGE, severe rotavirus gastroenteritis (SRVGE) and very severe rotavirus gastroenteritis (VSRVGE) were VE RVGE=35% (95% CI: 28%-41%), VE SRVGE=51% (95% CI: 33%-65%) and VE VSRVGE=64% (95% CI: 41%-78%). The protection rates of SRVGE in Asia and Africa are VE SRVGE=43% (95% CI: 28%-55%) and VE SRVGE=57% (95% CI: 17%-77%), respectively. There was no significant difference in the incidences of serious adverse events (SAEs) between RotaTeq vaccine group and placebo group ( χ2=2.05, P=0.152). Conclusions:RotaTeq vaccine has a certain protective effect on severe and above RVGE with good safety in high mortality areas in the world.

Objective:To investigate the epidemiological and clinical characteristics of G2P[4] group A rotavirus (RVA) in hospitalized children with diarrhea in China from 2016 to 2019, and to provide data support for the prevention and control of G2P[4] RVA.Methods:The data of viral diarrhea surveillance network in China from January 2016 to December 2019 were collected. A total of 19 667 specimens of hospitalized children with diarrhea under 5 years old were collected from all monitoring provinces, including 5 437 RVA positive specimens. EpiData 3.0 software and Excel 2010 were used for data collection and collation of viral diarrhea monitoring network, and SPSS 26.0 software was used for data analysis.Results:200 G2P[4] RVA specimens were detected from 5 437 RVA positive specimens, and the constituent rate of G2P[4] RVA was 3.68% (200/5 437) There is a statistically significant difference in the constituent ratio of G2P [4] RVA among RVA positive children in different years ( χ2=38.35, P<0.001), months ( χ2=62.69, P<0.001), and ages ( χ2=9.53, P=0.049). There is a statistically significant difference in the constituent ratio of G2P [4] RVA between rural and urban RVA positive children ( χ2=4.01, P=0.045). Compared with non-G2P[4] RVA hospitalized children, G2P[4] RVA hospitalized children had less proportion of respiratory tract infection ( χ2=6.07, P=0.014), G2P[4] RVA hospitalized children had higher proportion of fever ( χ2=6.68, P=0.010), there was no significant differences in diarrhea ( χ2=0.88, P=0.643), vomiting ( χ2=0.23, P=0.629), extraintestinal neurological symptoms ( χ2=0.18, P=0.668), and no significant difference in rash, sepsis and other complications ( χ2=0.45, P=0.504). Conclusions:The epidemic trend of G2P[4] RVA in China gradually decreased from 2016 to 2019, and the autumn and winter were G2P[4] RVA seasonal peaks. And the peak age was 24-36 months. There were a higher infection risk in rural areas, and fever was more than other genotypes.