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

This paper summarized the clinical experience of Prof. LI Diangui in treating gastroesophageal reflux disease （GERD） based on the theory of turbidity-toxin in liver. It is believed that internal accumulation of turbidity-toxin and liver depression with stomach counterflow are the main pathogenesis of GERD， and thus the therapeutic methods of resolving turbidity and resolving toxins， regulating the liver and harmonizing the stomach are proposed. In clinical practice， GERD is divided into the early stage， middle stage and late stage. For the early stage， the modified Huazhuo Shugan Hewei Formula （化浊疏肝和胃方） is used to regulate qi and remove turbidity， soothe the liver and harmonize the stomach； for the middle stage， the modified Huazhuo Qingre Zhisuan Formula （化浊清热制酸方） is applied to clear heat， direct the turbid downward， and resolve toxins； for the late stage， the modified Yiwei Decoction （益胃汤） is adopted to replenish qi， nourish yin and simultaneously resolve turbidity-toxin. Throughout the treatment process， attention should be paid to protecting the spleen and stomach， and the medication could be modified according to changes of individual condition.

Recently,British Society of Gastroenterology has published the guidelines on the management of autoimmune hepatitis(AIH).The guidelines provide guidance for the diagnosis and management of patients with AIH and overlap syndrome by reviewing and summarizing existing clinical evidence.The guidelines mainly focus on AIH in adults,but it is also applicable to adolescents and older children.This article presents an excerpt of the recommendations from the guidelines.

Objective:To compare the efficacy of robot system-assisted versus freehand screw revision for ankylosing spondylitis (AS) with lower cervical fractures.Methods:A multicenter retrospective cohort study was conducted to analyze the clinical data of 57 patients with AS combined with lower cervical fractures admitted to Honghui Hospital Affiliated to Xi'an Jiaotong University School of Medicine, Henan Provincial People's Hospital, Zhengzhou Orthopedic Hospital, and Qilu Hospital of Shandong University, including 46 males and 11 females, aged 38-77 years [(65.4±9.5)years]. Injury segments involved C 3 in 7 patients, C 4 in 13, C 5 in 25, C 6 in 10, and C 7 in 2. All the patients underwent revision surgery, among whom, 22 patients were treated with robot system-assisted cervical pedicle screw placement (robot nailing group, with 190 screws), and 35 with freehand cervical pedicle screw placement (freehand nailing group, with 300 screws). The operative duration, intraoperative bleeding volume, frequency of intraoperative fluoroscopy, incision length, and length of hospital stay of the two groups were compared; the time of single nscrew insertion, the number of single nail revisions, the distance between screws and the anterior cortex, the accuracy of screw placement of grade 0 and grade 0+1 were recorded in the two groups. The visual analogue scale (VAS), Japanese Orthopedic Society (JOA) score, neck dysfunction index (NDI), American Spine Injury Association (ASIA) classification before operation, at 3 days, 3 months after operation and at the last follow-up were compared between the two groups. The complication rate was also noted. Results:All the patients were followed up for 12-16 months [(14.3±2.1)months]. The operative duration, intraoperative bleeding volume, and frequency of intraoperative fluoroscopy were (186.4±12.9)minutes, (486.1±68.6)ml, and (3.4±1.3)times in the robot nailing group, which were shorter or less than (206.7±14.4)minutes, (660.3±45.2)ml, and (13.5±3.6)times in the freehand nailing group ( P<0.01). The incision length was (9.4±2.4)cm in the robot nailing group, longer than (5.6±1.2)cm in the freehand nailing group ( P<0.01), and the length of hospital stay was (3.7±0.4)days, shorter than (4.4±1.4)days in the freehand nailing group ( P<0.01). The length of single nail insertion, the number of single nail revision, and the distance between the screws and the front cortex were (6.5±0.4)minutes, (1.1±0.1)times, and (3.5±1.3)mm in the robot nailing group, which were shorter or less than (11.6±0.2)minutes, (1.5±0.2)times, and (12.4±4.7)mm in the freehand nailing group ( P<0.01). The accuracy of the screw placement in the robot nailing group was 90.0% (171/190) and 95.8% (182/190) with level 0 and 0+1 screws, better than 80.0% (240/300) and 89.0% (267/300) in the freehand nailing group ( P<0.05). There was no significant difference in VAS, JOA score, NDI, or ASIA grading between the two groups before operation ( P>0.05). The VAS, JOA, and NDI scores at 3 days after operation were (3.1±0.6)points, (12.1±1.2)points, and (15.6±2.9)points, respectively in the robot nailing group, which were better than (5.0±1.4)points, (11.3±1.1)points and (22.5±3.7)points, respectively in the freehand nailing group ( P<0.05). No statistically significant difference was observed in the ASIA grade between the two groups at 3 days after operation ( P>0.05). There were no significant differences in VAS, JOA, NDI scores, or ASIA grading between the two groups at 3 months after operation and at the last follow-up ( P>0.05). Compared with those before operation, the VAS, JOA, NDI scores, and ASIA grading were significantly improved at 3 days, 3 months after operation and at the last follow-up in the two groups, which were further improved with the passage of time. Two patients in the robot nailing group had pneumonia, with a complication rate of 9% (2/22), while 2 patients in the freehand nailing group had dural sac rupture and cerebrospinal fluid leakage and 3 had lung infection after operation, with a complication rate of 14% (5/35) ( P<0.05). Conclusion:Compared with freehand nailing, the robot system-assisted nailing revision for AS with lower cervical fracture has more advantages in terms of the operative duration, length of hospital stay, intraoperative bleeding volume, frequency of intraoperative fluoroscopy nailing speed and accuracy, screw holding force, early pain relief, function restoration, and complication rate, despite longer surgical incision.

Objective:To evaluate the quality of information on autoimmune liver disease in videos on the TikTok short video platform.Methods:The keyword "autoimmune liver disease" was used to search the top 200 videos on TikTok in the default sorting order. Using the DISCERN video quality assessment tool and the structured content integrity evaluation tool, we assessed the quality of the information in each video in relation to the pertinent disease guidelines. Furthermore, we investigated any relationships between the quality of the videos and their characteristics (likes, comments, retweets, days and duration of uploading).Results:A total of 140 videos were included, 96.4% of which were provided by medical professionals. The content completeness scores for each dimension were as follows: definition, 1.0 (0.0, 1.0); symptoms, 0.0 (0.0, 1.0); risk factors, 0.0 (0.0, 0.5); assessment, 0.5 (0.0, 1.5); management, 0.5 (0.0, 1.0); and outcome, 0.0 (0.0, 1.0). Furthermore, 91.4% of videos with DISCERN scores of ≤50 were of "fair" quality or below. Additionally, the difference in DISCERN scores between videos from different publishers was not statistically significant ( P>0.05). The number of likes, comments, favorites, retweets, and video duration had a strong positive correlation with the overall DISCERN score ( r=0.17, 0.18, 0.25, 0.26, 0.44, all P<0.05). Conclusions:The overall quality of videos related to autoimmune liver disease on the TikTok video platform is low. Therefore, publishers should focus on the comprehensiveness and accuracy of the information. Additionally, the TikTok platform should optimize its video review mechanism to provide the public with more accurate and reliable health information.

Recently,British Society of Gastroenterology has published the guidelines on the management of autoimmune hepatitis(AIH).The guidelines provide guidance for the diagnosis and management of patients with AIH and overlap syndrome by reviewing and summarizing existing clinical evidence.The guidelines mainly focus on AIH in adults,but it is also applicable to adolescents and older children.This article presents an excerpt of the recommendations from the guidelines.

Objective:To compare the efficacy of robot system-assisted versus freehand screw revision for ankylosing spondylitis (AS) with lower cervical fractures.Methods:A multicenter retrospective cohort study was conducted to analyze the clinical data of 57 patients with AS combined with lower cervical fractures admitted to Honghui Hospital Affiliated to Xi'an Jiaotong University School of Medicine, Henan Provincial People's Hospital, Zhengzhou Orthopedic Hospital, and Qilu Hospital of Shandong University, including 46 males and 11 females, aged 38-77 years [(65.4±9.5)years]. Injury segments involved C 3 in 7 patients, C 4 in 13, C 5 in 25, C 6 in 10, and C 7 in 2. All the patients underwent revision surgery, among whom, 22 patients were treated with robot system-assisted cervical pedicle screw placement (robot nailing group, with 190 screws), and 35 with freehand cervical pedicle screw placement (freehand nailing group, with 300 screws). The operative duration, intraoperative bleeding volume, frequency of intraoperative fluoroscopy, incision length, and length of hospital stay of the two groups were compared; the time of single nscrew insertion, the number of single nail revisions, the distance between screws and the anterior cortex, the accuracy of screw placement of grade 0 and grade 0+1 were recorded in the two groups. The visual analogue scale (VAS), Japanese Orthopedic Society (JOA) score, neck dysfunction index (NDI), American Spine Injury Association (ASIA) classification before operation, at 3 days, 3 months after operation and at the last follow-up were compared between the two groups. The complication rate was also noted. Results:All the patients were followed up for 12-16 months [(14.3±2.1)months]. The operative duration, intraoperative bleeding volume, and frequency of intraoperative fluoroscopy were (186.4±12.9)minutes, (486.1±68.6)ml, and (3.4±1.3)times in the robot nailing group, which were shorter or less than (206.7±14.4)minutes, (660.3±45.2)ml, and (13.5±3.6)times in the freehand nailing group ( P<0.01). The incision length was (9.4±2.4)cm in the robot nailing group, longer than (5.6±1.2)cm in the freehand nailing group ( P<0.01), and the length of hospital stay was (3.7±0.4)days, shorter than (4.4±1.4)days in the freehand nailing group ( P<0.01). The length of single nail insertion, the number of single nail revision, and the distance between the screws and the front cortex were (6.5±0.4)minutes, (1.1±0.1)times, and (3.5±1.3)mm in the robot nailing group, which were shorter or less than (11.6±0.2)minutes, (1.5±0.2)times, and (12.4±4.7)mm in the freehand nailing group ( P<0.01). The accuracy of the screw placement in the robot nailing group was 90.0% (171/190) and 95.8% (182/190) with level 0 and 0+1 screws, better than 80.0% (240/300) and 89.0% (267/300) in the freehand nailing group ( P<0.05). There was no significant difference in VAS, JOA score, NDI, or ASIA grading between the two groups before operation ( P>0.05). The VAS, JOA, and NDI scores at 3 days after operation were (3.1±0.6)points, (12.1±1.2)points, and (15.6±2.9)points, respectively in the robot nailing group, which were better than (5.0±1.4)points, (11.3±1.1)points and (22.5±3.7)points, respectively in the freehand nailing group ( P<0.05). No statistically significant difference was observed in the ASIA grade between the two groups at 3 days after operation ( P>0.05). There were no significant differences in VAS, JOA, NDI scores, or ASIA grading between the two groups at 3 months after operation and at the last follow-up ( P>0.05). Compared with those before operation, the VAS, JOA, NDI scores, and ASIA grading were significantly improved at 3 days, 3 months after operation and at the last follow-up in the two groups, which were further improved with the passage of time. Two patients in the robot nailing group had pneumonia, with a complication rate of 9% (2/22), while 2 patients in the freehand nailing group had dural sac rupture and cerebrospinal fluid leakage and 3 had lung infection after operation, with a complication rate of 14% (5/35) ( P<0.05). Conclusion:Compared with freehand nailing, the robot system-assisted nailing revision for AS with lower cervical fracture has more advantages in terms of the operative duration, length of hospital stay, intraoperative bleeding volume, frequency of intraoperative fluoroscopy nailing speed and accuracy, screw holding force, early pain relief, function restoration, and complication rate, despite longer surgical incision.

Objective:To evaluate the quality of information on autoimmune liver disease in videos on the TikTok short video platform.Methods:The keyword "autoimmune liver disease" was used to search the top 200 videos on TikTok in the default sorting order. Using the DISCERN video quality assessment tool and the structured content integrity evaluation tool, we assessed the quality of the information in each video in relation to the pertinent disease guidelines. Furthermore, we investigated any relationships between the quality of the videos and their characteristics (likes, comments, retweets, days and duration of uploading).Results:A total of 140 videos were included, 96.4% of which were provided by medical professionals. The content completeness scores for each dimension were as follows: definition, 1.0 (0.0, 1.0); symptoms, 0.0 (0.0, 1.0); risk factors, 0.0 (0.0, 0.5); assessment, 0.5 (0.0, 1.5); management, 0.5 (0.0, 1.0); and outcome, 0.0 (0.0, 1.0). Furthermore, 91.4% of videos with DISCERN scores of ≤50 were of "fair" quality or below. Additionally, the difference in DISCERN scores between videos from different publishers was not statistically significant ( P>0.05). The number of likes, comments, favorites, retweets, and video duration had a strong positive correlation with the overall DISCERN score ( r=0.17, 0.18, 0.25, 0.26, 0.44, all P<0.05). Conclusions:The overall quality of videos related to autoimmune liver disease on the TikTok video platform is low. Therefore, publishers should focus on the comprehensiveness and accuracy of the information. Additionally, the TikTok platform should optimize its video review mechanism to provide the public with more accurate and reliable health information.

Objective To train an automatic recognition and classification model of ulcerative colitis colonoscopy image based on ResNet,and to test its accuracy,in order to help doctors improve the clinical detection rate and classification accuracy of ulcerative colitis.Methods A total of 4000 colonoscopy images were retrospectively collected from the Colonoscopy Center of Hebei Hospital of Traditional Chinese Medicine from January 2018 to October 2023,and were divided into normal group,mild group,moderate group and severe group according to Mayo endoscopic scoring criteria,with 1000 images for each group.After pre-processing such as brightness adjustment and Angle rotation,the number of images was expanded to 20,000,and the data set was randomly divided into training set,verification set and test set according to the ratio of 7∶2∶1.The training set and verification set are input into the ResNet model to learn and test its stability.After all training is completed,the accuracy of the model is recorded through the test set,and the accurate regression curve is made to evaluate the classification effect of the model.Results In the test set,the accuracy of classification of ulcerative colitis was 99.8％in normal group,98.8％in mild group,95.6％in moderate group and 97.8％in severe group.Conclusion ResNet has good performance in image recognition and classification of ulcerative colitis,can improve the clinical accuracy of ulcerative colitis,and can assist doctors to identify and classify the disease,which has a more reliable clinical application value.