1.Guideline-driven clinical decision support for colonoscopy patients using the hierarchical multi-label deep learning method.
Junling WU ; Jun CHEN ; Hanwen ZHANG ; Zhe LUAN ; Yiming ZHAO ; Mengxuan SUN ; Shufang WANG ; Congyong LI ; Zhizhuang ZHAO ; Wei ZHANG ; Yi CHEN ; Jiaqi ZHANG ; Yansheng LI ; Kejia LIU ; Jinghao NIU ; Gang SUN
Chinese Medical Journal 2025;138(20):2631-2639
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
2.Fabrication and biological effect investigation of the antibacterial drug delivery system with PEG-hydrogel and gentamicin-loaded-CSt on titanium surface
Yunfeng WU ; Litao LI ; Zhizhuang ZHAO ; Long YU ; Yunlong MA ; Dan LI ; Dawei LI ; Decheng WU ; Fei YANG ; Yuanzheng MA
Chinese Journal of Orthopaedics 2018;38(12):731-741
Objective To fabricate an antibacterial controlled drug delivery system with PEG-hydrogel and gentamicin-loaded-CSt on titanium surface,and to investigate its surface characteristics,swelling behavior,drug release behavior in vitro,antiinfection performance in vivo,and tissue biocompatibility.Methods Cross-linked starch (CSt) was synthesized first and then CSt was loaded with gentamicin (GEN) as a carrier (GEN@CSt),then 4-arm-polyethylene glycol (PEG) was added to it which was mixed by ultrasound.The surface of titanium (Ti) was covered with a layer of poly dopamine (PDA).The drug-loaded hydrogel was fixed to the titanium surface,subsequently capped by poly lactic-co-glycolic acid (PLGA) membranes,and then the Ti-PDA-PEG (GEN@CSt)-PLGA composite coating was fabricated finally.Surface morphology of the system was observed,while the swelling behavior was characterized;release behavior of the composite coating was detected;the bacteriostatic experiments were carried out with staphylococcus aureus (SAU),staphylococcus epidermidis (SEP) and escherichia coli (ECO) in vitro.The animal models of infected bone defect was established in 36 New Zealand white rabbits.These animals were randomly divided into three groups.Group 1 animals were implanted with drug-loaded composite coatings.Group 2 animals were implanted with drug-free composite coatings.Group 3 animals were implanted with bare titanium rods.The infection data were collected periodically to carry out antiinfection experiments in vivo.Another 12 rabbits were divided into the experimental group and the control group randomly.Biocompatibility of the materials was observed by histopathology after implantation of the corresponding materials into the femoral condyle.Results The composite coating adhered to the titanium surface firmly,presenting a smooth and translucent shape.The ratio of CSt/PEG affects swelling behavior varied,starch-free gels maintained an equilibrium swelling of 7.4,after the ratio reached 1 ∶ 1,the equilibrium swelling ratio remained at 3.0.In-vitro the release rate of the first 8 h was fast,and the cumulative release amount accounted for 83% of the total in the first 7 days,lasting more than 13 d.In vitro antibacterial test,the average diameter of the inhibition ring was 3.6±0.13 cm (SAU),3.4±0.11 cm (SEP),3.7±0.10 cm (ECO).In-vivo anti-infection experiment,the infection situation of the group 1 was better than the control groups 2 and 3.The pathological results indicated that inflammatory reaction in the experimental group was basically the same as the control group.Conclusion The study successfully fabricated the antibacterial controlled drug delivery system with PEG-hydrogel and gentamicin-loaded-CSt on titanium surface.The system has a reasonable drug release behavior,and effectively inhibited the growth of bacteria in vivo and in vitro.It also has good biocompatibility to stand a promising strategy to improve the orthopedics anti-infection.

Result Analysis
Print
Save
E-mail