ObjectiveTo construct a large language model （LLM）-based intelligent pre-consultation system for traditional Chinese medicine （TCM） to improve efficacy of clinical practice. MethodsA TCM large language model was fine-tuned using DeepSpeed ZeRO-3 distributed training strategy based on YAYI 2-30B. A weighted undirected graph network was designed and an agent-based syndrome differentiation model was established based on relationship data extracted from TCM literature and clinical records. An agent collaboration framework was developed to integrate the TCM LLM with the syndrome differentiation model. Model performance was comprehensively evaluated by Loss function， BLEU-4， and ROUGE-L metrics， through which training convergence， text generation quality， and language understanding capability were assessed. Professional knowledge test sets were developed to evaluate system proficiency in TCM physician licensure content， TCM pharmacist licensure content， TCM symptom terminology recognition， and meridian identification. Clinical tests were conducted to compare the system with attending physicians in terms of diagnostic accuracy， consultation rounds， and consultation duration. ResultsAfter 100 000 iterations， the training loss value was gradually stabilized at about 0.7±0.08， indicating that the TCM-LLM has been trained and has good generalization ability. The TCM-LLM scored 0.38 in BLEU-4 and 0.62 in ROUGE-L， suggesting that its natural language processing ability meets the standard. We obtained 2715 symptom terms， 505 relationships between diseases and syndromes， 1011 relationships between diseases and main symptoms， and 1 303 600 relationships among different symptoms， and constructed the Agent of syndrome differentiation model. The accuracy rates in the simulated tests for TCM practitioners， licensed pharmacists of Chinese materia medica， recognition of TCM symptom terminology， and meridian recognition were 94.09%， 78.00%， 87.50%， and 68.80%， respectively. In clinical tests， the syndrome differentiation accuracy of the system reached 88.33%， with fewer consultation rounds and shorter consultation time compared to the attending physicians （P＜0.01）， suggesting that the system has a certain pre- consultation ability. ConclusionThe LLM-based intelligent TCM pre-diagnosis system could simulate diagnostic thinking of TCM physicians to a certain extent. After understanding the patients' natural language， it collects all the patient's symptom through guided questioning， thereby enhancing the diagnostic and treatment efficiency of physicians as well as the consultation experience of the patients.

Objective The fully automatic segmentation of glioma and its subregions is fundamental for computer-aided clinical diagnosis of tumors.In the segmentation process of brain magnetic resonance imaging(MRI),convolutional neural networks with small convolutional kernels can only capture local features and are ineffective at integrating global features,which narrows the receptive field and leads to insufficient segmentation accuracy.This study aims to use dilated convolution to address the problem of inadequate global feature extraction in 3D-UNet.Methods 1)Algorithm construction:A 3D-UNet model with three pathways for more global contextual feature extraction,or 3DGE-UNet,was proposed in the paper.By using publicly available datasets from the Brain Tumor Segmentation Challenge(BraTS)of 2019(335 patient cases),a global contextual feature extraction(GE)module was designed.This module was integrated at the first,second,and third skip connections of the 3D UNet network.The module was utilized to fully extract global features at different scales from the images.The global features thus extracted were then overlaid with the upsampled feature maps to expand the model's receptive field and achieve deep fusion of features at different scales,thereby facilitating end-to-end automatic segmentation of brain tumors.2)Algorithm validation:The image data were sourced from the BraTs 2019 dataset,which included the preoperative MRI images of 335 patients across four modalities(T1,T1ce,T2,and FLAIR)and a tumor image with annotations made by physicians.The dataset was divided into the training,the validation,and the testing sets at an 8∶1∶1 ratio.Physician-labelled tumor images were used as the gold standard.Then,the algorithm's segmentation performance on the whole tumor(WT),tumor core(TC),and enhancing tumor(ET)was evaluated in the test set using the Dice coefficient(for overall effectiveness evaluation),sensitivity(detection rate of lesion areas),and 95%Hausdorff distance(segmentation accuracy of tumor boundaries).The performance was tested using both the 3D-UNet model without the GE module and the 3DGE-UNet model with the GE module to internally validate the effectiveness of the GE module setup.Additionally,the performance indicators were evaluated using the 3DGE-UNet model,ResUNet,UNet++,nnUNet,and UNETR,and the convergence of these five algorithm models was compared to externally validate the effectiveness of the 3DGE-UNet model.Results 1)In internal validation,the enhanced 3DGE-UNet model achieved Dice mean values of 91.47%,87.14%,and 83.35%for segmenting the WT,TC,and ET regions in the test set,respectively,producing the optimal values for comprehensive evaluation.These scores were superior to the corresponding scores of the traditional 3D-UNet model,which were 89.79%,85.13%,and 80.90%,indicating a significant improvement in segmentation accuracy across all three regions(P＜0.05).Compared with the 3D-UNet model,the 3DGE-UNet model demonstrated higher sensitivity for ET(86.46%vs.80.77%)(P＜0.05),demonstrating better performance in the detection of all the lesion areas.When dealing with lesion areas,the 3DGE-UNet model tended to correctly identify and capture the positive areas in a more comprehensive way,thereby effectively reducing the likelihood of missed diagnoses.The 3DGE-UNet model also exhibited exceptional performance in segmenting the edges of WT,producing a mean 95%Hausdorff distance superior to that of the 3D-UNet model(8.17 mm vs.13.61 mm,P＜0.05).However,its performance for TC(8.73 mm vs.7.47 mm)and ET(6.21 mm vs.5.45 mm)was similar to that of the 3D-UNet model.2)In the external validation,the other four algorithms outperformed the 3DGE-UNet model only in the mean Dice for TC(87.25%),the mean sensitivity for WT(94.59%),the mean sensitivity for TC(86.98%),and the mean 95%Hausdorff distance for ET(5.37 mm).Nonetheless,these differences were not statistically significant(P＞0.05).The 3DGE-UNet model demonstrated rapid convergence during the training phase,outpacing the other external models.Conclusion The 3DGE-UNet model can effectively extract and fuse feature information on different scales,improving the accuracy of brain tumor segmentation.

Objective: To analyze the ear, nose, and throat exam of some freshmen in the military college entrance examination in Shandong Province in 2020 and to facilitate adolescent targeted health promotion. Methods: The 1 411 freshmen participating in the military college entrance examination in Jinan, Zibo and Weifang of Shandong Province were included. The ear, nose, and throat exam were performed by professionals using electric otoscope, 5 meter whispering test, and front rhinoscope. Results: Nasal septal deviation and hypertrophy of inferior turbinate accounted for the highest proportion. Among 489 cases of nasal septum deviation, the detection rate of Jinan (15.97%) was significantly lower than that of Weifang (43.60%) and Zibo (46.53%) ( χ 2=63.32， P <0.05). For deviation of nasal septum, the detection rate in students with urban residence (31.53%) was significantly lower than that of rural students (39.03%) ( χ 2=4.11， P <0.05). Seventy two cases of inferior turbinate hyperplasia were detected, and the detection rate in Jinan (2.99%) was significantly lower than that in Weifang (6.51%) and Zibo (6.04%) ( χ 2=6.63， P <0.05). The detection rate of tonsil hypertrophy was significantly lower in boys (4.63%), students from urban area (3.94%), compared with that of girls(9.56%) and rural students (6.70%) ( χ 2=5.35，4.86， P <0.05). In pharyngeal examination, tonsil hyperplasia was the most common condition of enlarged tonsils ( n =214), which was significantly higher in Jinan(22.36%) than that of Weifang (11.71 %) and Zibo (10.74%) ( χ 2=22.39， P <0.05), and was significantly lower in boys (14.38%) and rural students (12.40%) than that in girls (22.79%) and urban students (17.24%) ( χ 2=4.70，4.65， P <0.05). Conclusion Nasal septum deviation and tonsil hypertrophy are the most prevalent upper airway diseases among freshmen participating in the military college entrance examination. Prevention and treatment of nasopharynx diseases should be emphasized.