Objective:To evaluate the diagnostic value of the vesical imaging-reporting and data system（VI-RADS）for determining muscle invasion in variant histology urothelial carcinoma（VUC）of the bladder.Methods:A retrospective analysis was performed on the pathological and imaging data of 518 bladder cancer patients admitted to Jiangsu Province Hospital between January 2013 and January 2023. Patients were stratified into pure urothelial carcinoma（PUC）group（ n = 457）and variant urothelial carcinoma（VUC）group（ n = 61）based on the presence of histological variants. In the PUC group，there were 390 males（85.3%）and 67 females（14.7%），with a mean age of（66.9 ± 11.2）years. Tumor characteristics included maximum diameter ≥ 30 mm in 149（32.6%），< 30 mm in 308（67.4%），multiple tumors in 147（32.2%），solitary in 310（67.8%），pedunculated morphology in 143（31.3%）and non-pedunculated in 314（68.7%）. Histological grading identified high-grade tumors in 319 patients（69.8%）and low-grade tumors in 138（30.2%）. Pathological stage distribution included 191 of T a（41.8%），127 of T 1（27.8%），76 of T 2（16.6%），47 of T 3（10.3%），and 16 of T 4（3.5%）patients. The VUC group included 61 patients，comprising 51 males（83.6%）and 10 females（16.4%），with a mean age of（65.8 ± 11.4）years. Tumor characteristics were maximum diameter ≥ 30 mm in 38（62.3%），< 30 mm in 23（37.7%），multiple tumors in 16（26.2%），solitary in 45（73.8%），pedunculated morphology in 11（18.0%）and non-pedunculated in 50（82.0%）. Histological grading identified high-grade tumors in 59 patients（96.7%）and low-grade tumors in 2（3.3%）. Pathological stage distribution included 3 of T a（4.9%），15 of T 1（24.6%），15 of T 2（24.6%），20 of T 3（32.8%），and 8 of T 4（13.1%）patients. No statistically significant differences were found between the two groups in gender，age，or tumor multiplicity（ P > 0.05）. Statistically significant differences were found in pathological grade，pathological stage，maximum tumor diameter，and pedunculated morphology（ P < 0.05）. Furthermore，an external validation cohort of 278 bladder cancer patients treated between February 2023 and February 2024 from multiple centers（Jiangsu Provincial People’s Hospital，The First Affiliated Hospital of Zhengzhou University，Union Hospital Tongji Medical College Huazhong University of Science and Technology，Jiangsu Provincial Hospital of Traditional Chinese Medicine，Suzhou Municipal Hospital，Huaian First People’s Hospital，Yixing People’s Hospital）was retrospectively analyzed to externally validate the performance of VI-RADS scoring in predicting muscle invasion of VUC. This cohort included a PUC subgroup of 241 patients，comprising 196 males（81.3%）and 45 females（18.7%），with a mean age of（68.0 ± 10.7）years. Tumor characteristics were maximum diameter ≥ 30 mm in 85（35.3%），< 30 mm in 156（64.7%），multiple tumors in 65（27.0%），solitary in 176（73.0%），pedunculated morphology in 76（31.5%）and non-pedunculated in 165（68.5%）. Histological grading identified high-grade tumors in 175 patients（72.6%）and low-grade tumors in 66（27.4%）. Pathological staging comprised 107 patients of T a（44.4%），78 of T 1（32.4%），22 of T 2（9.1%），22 of T 3（9.1%），and 12 of T 4（5.0%）. The VUC subgroup consisted of 37 patients，comprising 29 males（78.4%）and 8 females（21.6%），with a mean age of（70.5 ± 9.5）years. Tumor characteristics were maximum diameter ≥ 30 mm in 23（62.2%），< 30 mm in 14（37.8%），multiple tumors in 9（24.3%），solitary in 28（75.7%），pedunculated morphology in 7（18.9%）and non-pedunculated in 30（81.1%）. Histological grading identified high-grade tumors in 36 patients（97.3%）and low-grade tumors in 1（2.7%）. Pathological staging comprised 1 patient of T a（2.7%），9 of T 1（24.3%），7 of T 2（18.9%），19 of T 3（51.4%），and 1 of T 4（2.7%）. In this validation cohort，no significant differences were found in gender，age，tumor multiplicity，or pedunculated morphology between the PUC and VUC subgroups（ P > 0.05）. Significant differences were observed in pathological grade，pathological stage，and maximum tumor diameter（ P < 0.05）. Three radiologists independently reviewed and scored the multiparametric MRI（mp-MRI）in a blinded manner. Inter-reader agreement was assessed using the weighted kappa statistic. Differences in variables between the two groups were compared using t-tests，chi-square tests，or Fisher’s exact test. The diagnostic performance of VI-RADS for muscle invasion in VUC and PUC was comprehensively evaluated using receiver operating characteristic（ROC）curves，the area under the curve（AUC），and cut-off values determined by the Youden’s index. The DeLong test was used to assess whether the diagnostic performance of VI-RADS differed between VUC and PUC. Results:In the retrospective single-center cohort，the AUC of VI-RADS for assessing muscle invasion was 0.895（95% CI 0.864?0.922）in the PUC group，with a cut-off value of > 3，and the AUC was 0.896（95% CI 0.791-0.960）in the VUC group，with a cut-off value of > 3. The difference between the two groups was not statistically significant（ P = 0.986）. Using a VI-RADS score > 3 as the cut-off value，the accuracy，sensitivity，specificity，positive predictive value（PPV），and negative predictive value（NPV）for diagnosing muscle invasion status in the PUC group were 85.8%（392/457），70.5%（98/139），92.5%（294/318），80.3%（98/122），and 87.8%（294/335），respectively. The corresponding values for the VUC group were 82.0%（50/61），76.7%（33/43），94.4%（17/18），97.1%（33/34），and 63.0%（17/27）.In the retrospective multicenter cohort，the AUC of VI-RADS for assessing muscle invasion was 0.891（95% CI 0.845?0.927）in the PUC group，with a cut-off value of > 2，and the AUC was 0.898（95% CI 0.754?0.973）in the VUC group，with a cut-off value of > 3. The difference between the two groups was not statistically significant（ P = 0.897）. Using a VI-RADS score > 3 as the cut-off value，the accuracy，sensitivity，specificity，PPV，and NPV for diagnosing muscle invasion status in the PUC group were 85.9%（207/241），58.9%（33/56），94.1%（174/185），75.0%（33/44），and 88.3%（174/197），respectively. The corresponding values for the VUC group were 81.1%（30/37），77.8%（21/27），90.0%（9/10），95.5%（21/22），and 60.0%（9/15）.In the single-center cohort，the Kappa values for inter-reader agreement in assessing muscle invasion status using VI-RADS were 0.881（ P < 0.01）for the PUC group and 0.941（ P < 0.01）for the VUC group among the three readers. In the multicenter cohort，the Kappa values were 0.858（ P < 0.01）for the PUC group and 0.838（ P < 0.01）for the VUC group. Conclusions:VI-RADS demonstrates similarly high diagnostic performance for assessing muscle invasion in both PUC and VUC，which is applicable for diagnosing muscle invasion status in VUC，and shows good inter-reader agreement.

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.

Esophageal cancer is the eighth most common malignant tumor in the world, and China is a country with high incidence of esophageal cancer. Currently, The treatment options for resectable esophageal cancer is multidisciplinary comprehensive treatment based on surgery. In recent years, minimally invasive esophageal cancer surgery has been increasingly used in clinical practice, especially McKeown minimally invasive esophagectomy(MIE) has become the most minimally invasive esophagectomy currently used in China. But how effective is the McKeown MIE for esophageal cancer compared to the open McKeown and other minimally invasive esophageal cancer procedures? A detailed comprehensive description is not yet available. In this paper, We will summarize the clinical studies related to McKeown MIE and traditional open McKeown, Ivor-Lewis MIE, Sweet operation and robot-assisted McKeown MIE, and systematically analyze them, so as to clarify the value of McKeown MIE surgery in the treatment of esophageal cancer.

Objective:To evaluate the diagnostic value of the vesical imaging-reporting and data system（VI-RADS）for determining muscle invasion in variant histology urothelial carcinoma（VUC）of the bladder.Methods:A retrospective analysis was performed on the pathological and imaging data of 518 bladder cancer patients admitted to Jiangsu Province Hospital between January 2013 and January 2023. Patients were stratified into pure urothelial carcinoma（PUC）group（ n = 457）and variant urothelial carcinoma（VUC）group（ n = 61）based on the presence of histological variants. In the PUC group，there were 390 males（85.3%）and 67 females（14.7%），with a mean age of（66.9 ± 11.2）years. Tumor characteristics included maximum diameter ≥ 30 mm in 149（32.6%），< 30 mm in 308（67.4%），multiple tumors in 147（32.2%），solitary in 310（67.8%），pedunculated morphology in 143（31.3%）and non-pedunculated in 314（68.7%）. Histological grading identified high-grade tumors in 319 patients（69.8%）and low-grade tumors in 138（30.2%）. Pathological stage distribution included 191 of T a（41.8%），127 of T 1（27.8%），76 of T 2（16.6%），47 of T 3（10.3%），and 16 of T 4（3.5%）patients. The VUC group included 61 patients，comprising 51 males（83.6%）and 10 females（16.4%），with a mean age of（65.8 ± 11.4）years. Tumor characteristics were maximum diameter ≥ 30 mm in 38（62.3%），< 30 mm in 23（37.7%），multiple tumors in 16（26.2%），solitary in 45（73.8%），pedunculated morphology in 11（18.0%）and non-pedunculated in 50（82.0%）. Histological grading identified high-grade tumors in 59 patients（96.7%）and low-grade tumors in 2（3.3%）. Pathological stage distribution included 3 of T a（4.9%），15 of T 1（24.6%），15 of T 2（24.6%），20 of T 3（32.8%），and 8 of T 4（13.1%）patients. No statistically significant differences were found between the two groups in gender，age，or tumor multiplicity（ P > 0.05）. Statistically significant differences were found in pathological grade，pathological stage，maximum tumor diameter，and pedunculated morphology（ P < 0.05）. Furthermore，an external validation cohort of 278 bladder cancer patients treated between February 2023 and February 2024 from multiple centers（Jiangsu Provincial People’s Hospital，The First Affiliated Hospital of Zhengzhou University，Union Hospital Tongji Medical College Huazhong University of Science and Technology，Jiangsu Provincial Hospital of Traditional Chinese Medicine，Suzhou Municipal Hospital，Huaian First People’s Hospital，Yixing People’s Hospital）was retrospectively analyzed to externally validate the performance of VI-RADS scoring in predicting muscle invasion of VUC. This cohort included a PUC subgroup of 241 patients，comprising 196 males（81.3%）and 45 females（18.7%），with a mean age of（68.0 ± 10.7）years. Tumor characteristics were maximum diameter ≥ 30 mm in 85（35.3%），< 30 mm in 156（64.7%），multiple tumors in 65（27.0%），solitary in 176（73.0%），pedunculated morphology in 76（31.5%）and non-pedunculated in 165（68.5%）. Histological grading identified high-grade tumors in 175 patients（72.6%）and low-grade tumors in 66（27.4%）. Pathological staging comprised 107 patients of T a（44.4%），78 of T 1（32.4%），22 of T 2（9.1%），22 of T 3（9.1%），and 12 of T 4（5.0%）. The VUC subgroup consisted of 37 patients，comprising 29 males（78.4%）and 8 females（21.6%），with a mean age of（70.5 ± 9.5）years. Tumor characteristics were maximum diameter ≥ 30 mm in 23（62.2%），< 30 mm in 14（37.8%），multiple tumors in 9（24.3%），solitary in 28（75.7%），pedunculated morphology in 7（18.9%）and non-pedunculated in 30（81.1%）. Histological grading identified high-grade tumors in 36 patients（97.3%）and low-grade tumors in 1（2.7%）. Pathological staging comprised 1 patient of T a（2.7%），9 of T 1（24.3%），7 of T 2（18.9%），19 of T 3（51.4%），and 1 of T 4（2.7%）. In this validation cohort，no significant differences were found in gender，age，tumor multiplicity，or pedunculated morphology between the PUC and VUC subgroups（ P > 0.05）. Significant differences were observed in pathological grade，pathological stage，and maximum tumor diameter（ P < 0.05）. Three radiologists independently reviewed and scored the multiparametric MRI（mp-MRI）in a blinded manner. Inter-reader agreement was assessed using the weighted kappa statistic. Differences in variables between the two groups were compared using t-tests，chi-square tests，or Fisher’s exact test. The diagnostic performance of VI-RADS for muscle invasion in VUC and PUC was comprehensively evaluated using receiver operating characteristic（ROC）curves，the area under the curve（AUC），and cut-off values determined by the Youden’s index. The DeLong test was used to assess whether the diagnostic performance of VI-RADS differed between VUC and PUC. Results:In the retrospective single-center cohort，the AUC of VI-RADS for assessing muscle invasion was 0.895（95% CI 0.864?0.922）in the PUC group，with a cut-off value of > 3，and the AUC was 0.896（95% CI 0.791-0.960）in the VUC group，with a cut-off value of > 3. The difference between the two groups was not statistically significant（ P = 0.986）. Using a VI-RADS score > 3 as the cut-off value，the accuracy，sensitivity，specificity，positive predictive value（PPV），and negative predictive value（NPV）for diagnosing muscle invasion status in the PUC group were 85.8%（392/457），70.5%（98/139），92.5%（294/318），80.3%（98/122），and 87.8%（294/335），respectively. The corresponding values for the VUC group were 82.0%（50/61），76.7%（33/43），94.4%（17/18），97.1%（33/34），and 63.0%（17/27）.In the retrospective multicenter cohort，the AUC of VI-RADS for assessing muscle invasion was 0.891（95% CI 0.845?0.927）in the PUC group，with a cut-off value of > 2，and the AUC was 0.898（95% CI 0.754?0.973）in the VUC group，with a cut-off value of > 3. The difference between the two groups was not statistically significant（ P = 0.897）. Using a VI-RADS score > 3 as the cut-off value，the accuracy，sensitivity，specificity，PPV，and NPV for diagnosing muscle invasion status in the PUC group were 85.9%（207/241），58.9%（33/56），94.1%（174/185），75.0%（33/44），and 88.3%（174/197），respectively. The corresponding values for the VUC group were 81.1%（30/37），77.8%（21/27），90.0%（9/10），95.5%（21/22），and 60.0%（9/15）.In the single-center cohort，the Kappa values for inter-reader agreement in assessing muscle invasion status using VI-RADS were 0.881（ P < 0.01）for the PUC group and 0.941（ P < 0.01）for the VUC group among the three readers. In the multicenter cohort，the Kappa values were 0.858（ P < 0.01）for the PUC group and 0.838（ P < 0.01）for the VUC group. Conclusions:VI-RADS demonstrates similarly high diagnostic performance for assessing muscle invasion in both PUC and VUC，which is applicable for diagnosing muscle invasion status in VUC，and shows good inter-reader agreement.

Esophageal cancer is the eighth most common malignant tumor in the world, and China is a country with high incidence of esophageal cancer. Currently, The treatment options for resectable esophageal cancer is multidisciplinary comprehensive treatment based on surgery. In recent years, minimally invasive esophageal cancer surgery has been increasingly used in clinical practice, especially McKeown minimally invasive esophagectomy(MIE) has become the most minimally invasive esophagectomy currently used in China. But how effective is the McKeown MIE for esophageal cancer compared to the open McKeown and other minimally invasive esophageal cancer procedures? A detailed comprehensive description is not yet available. In this paper, We will summarize the clinical studies related to McKeown MIE and traditional open McKeown, Ivor-Lewis MIE, Sweet operation and robot-assisted McKeown MIE, and systematically analyze them, so as to clarify the value of McKeown MIE surgery in the treatment of esophageal cancer.

Objective To investigate residents'willingness to use"internet+nursing services"of a community in Xiamen City and provide reference for the code of practice to promote the service.Methods A total of 350 community residents in Xiamen City were investigated by convenience sampling from January to December 2023.The questionnaire was self-designed with regard to their willingness to use"internet+nursing services".Results 336 questionnaires were effectively collected.The total score of willingness to use"internet+nursing services"of community residents in Xiamen City was(68.80±8.73)points,but only 19.6％of residents knew about"internet+nursing services".There were statistically significant differences in willingness to use among community residents with different ages,medical payment type,whether they knew"internet+nursing services"or not,and the number of service items needed(P＜0.05).Among them,age,whether they knew"internet+nursing services"or not and the number of service items needed were the main influencing factors of willingness to use(P＜0.05).Conclusion Community residents in Xiamen City show above the middle level of willingness to use"internet+nursing services".However,the popularity of"internet+nursing services"has a large room for improvement.The publicity to the residents should be strengthened,and the care needs of residents with different demographic characteristics should be considered.The supporting application functions of such platforms should be designed to meet the health needs of community residents,especially elderly patients with great demand for services.

Objective To investigate residents'willingness to use"internet+nursing services"of a community in Xiamen City and provide reference for the code of practice to promote the service.Methods A total of 350 community residents in Xiamen City were investigated by convenience sampling from January to December 2023.The questionnaire was self-designed with regard to their willingness to use"internet+nursing services".Results 336 questionnaires were effectively collected.The total score of willingness to use"internet+nursing services"of community residents in Xiamen City was(68.80±8.73)points,but only 19.6％of residents knew about"internet+nursing services".There were statistically significant differences in willingness to use among community residents with different ages,medical payment type,whether they knew"internet+nursing services"or not,and the number of service items needed(P＜0.05).Among them,age,whether they knew"internet+nursing services"or not and the number of service items needed were the main influencing factors of willingness to use(P＜0.05).Conclusion Community residents in Xiamen City show above the middle level of willingness to use"internet+nursing services".However,the popularity of"internet+nursing services"has a large room for improvement.The publicity to the residents should be strengthened,and the care needs of residents with different demographic characteristics should be considered.The supporting application functions of such platforms should be designed to meet the health needs of community residents,especially elderly patients with great demand for services.

Objective To evaluate the performance of magnetic resonance imaging(MRI)multi-sequence feature imputation and fusion mutual model based on sequence deletion in differentiating high-grade glioma(HGG)from low-grade glioma(LGG).Methods We retrospectively collected multi-sequence MR images from 305 glioma patients,including 189 HGG patients and 116 LGG patients.The region of interest(ROI)of T1-weighted images(T1WI),T2-weighted images(T2WI),T2 fluid attenuated inversion recovery(T2_FLAIR)and post-contrast enhancement T1WI(CE_T1WI)were delineated to extract the radiomics features.A mutual-aid model of MRI multi-sequence feature imputation and fusion based on sequence deletion was used for imputation and fusion of the feature matrix with missing data.The discriminative ability of the model was evaluated using 5-fold cross-validation method and by assessing the accuracy,balanced accuracy,area under the ROC curve(AUC),specificity,and sensitivity.The proposed model was quantitatively compared with other non-holonomic multimodal classification models for discriminating HGG and LGG.Class separability experiments were performed on the latent features learned by the proposed feature imputation and fusion methods to observe the classification effect of the samples in two-dimensional plane.Convergence experiments were used to verify the feasibility of the model.Results For differentiation of HGG from LGG with a missing rate of 10%,the proposed model achieved accuracy,balanced accuracy,AUC,specificity,and sensitivity of 0.777,0.768,0.826,0.754 and 0.780,respectively.The fused latent features showed excellent performance in the class separability experiment,and the algorithm could be iterated to convergence with superior classification performance over other methods at the missing rates of 30%and 50%.Conclusion The proposed model has excellent performance in classification task of HGG and LGG and outperforms other non-holonomic multimodal classification models,demonstrating its potential for efficient processing of non-holonomic multimodal data.

Objective To investigate the genetic causes of auditory neuropathy with optic atrophy in a family.Methods The proband's medical history and family history were inquired in detail,and relevant clinical examina-tions were performed to confirm the diagnosis of auditory neuropathy with optic atrophy,and the genetic pedigree of the family was drawn.Peripheral blood of proband(Ⅲ-7)was collected for whole exome sequencing,and the patho-genicity of the detected mutations were interpreted.Blood samples of proband's wife(Ⅲ-8),eldest daughter(Ⅳ-7),second daughter(Ⅳ-9)and son(Ⅳ-10)were tested for mutation sites by Sanger sequencing.Combined with clinical manifestations and examination results,the family was studied.Results The genetic pattern of this family was autosomal dominant.The proband showed decreased visual acuity at the age of 19,bilateral sensorineural deaf-ness at the age of 30,and decreased speech recognition rate.Among 20 members of the family of 5 generations,10(2 deceased)showed similar symptoms of hearing and visual impairment.Proband(Ⅲ-7),eldest daughter(Ⅳ-7)and son(Ⅳ-10)underwent relevant examination.Pure tone audiometry showed bilateral sensorineural deafness.ABR showed no response bilaterally.The 40 Hz AERP showed no response in both ears.OAE showed responses in some or all of the frequencies.No stapedial reflex was detected.The eye movement of Ⅲ-7 and Ⅳ-10 were reasona-ble in all directions,and color vision was normal.Ocular papilla atrophy was observed in different degrees in fundus examination.OCT showed thinning of optic disc nerve fibers in both eyes,and visual evoked potential showed pro-longed P100 wave peak.They were diagnosed as hereditary auditory neuropathy with optic atrophy.A mutation of the OPA1 gene c.1334G＞A(p.Arg445His,NM_015560.2)at a pathogenic locus on chromosome 3 was detected by whole exon detection in Ⅲ-7.The results of generation sequencing analysis showed that the OPA1 gene c.1334G＞A(p.Arg445His,NM_015560.2)mutation of chromosome 3 was also found in Ⅳ-7 and Ⅳ-10.Meanwhile,the gen-otypes of Ⅲ-8 and Ⅳ-9 were wild homozygous,that is,no mutation occurred.Conclusion The OPA1 c.1334G＞A(p.Arg445His,NM_015560.2)mutation site might be the pathogenic mutation in this family.

Objective To evaluate the performance of magnetic resonance imaging(MRI)multi-sequence feature imputation and fusion mutual model based on sequence deletion in differentiating high-grade glioma(HGG)from low-grade glioma(LGG).Methods We retrospectively collected multi-sequence MR images from 305 glioma patients,including 189 HGG patients and 116 LGG patients.The region of interest(ROI)of T1-weighted images(T1WI),T2-weighted images(T2WI),T2 fluid attenuated inversion recovery(T2_FLAIR)and post-contrast enhancement T1WI(CE_T1WI)were delineated to extract the radiomics features.A mutual-aid model of MRI multi-sequence feature imputation and fusion based on sequence deletion was used for imputation and fusion of the feature matrix with missing data.The discriminative ability of the model was evaluated using 5-fold cross-validation method and by assessing the accuracy,balanced accuracy,area under the ROC curve(AUC),specificity,and sensitivity.The proposed model was quantitatively compared with other non-holonomic multimodal classification models for discriminating HGG and LGG.Class separability experiments were performed on the latent features learned by the proposed feature imputation and fusion methods to observe the classification effect of the samples in two-dimensional plane.Convergence experiments were used to verify the feasibility of the model.Results For differentiation of HGG from LGG with a missing rate of 10%,the proposed model achieved accuracy,balanced accuracy,AUC,specificity,and sensitivity of 0.777,0.768,0.826,0.754 and 0.780,respectively.The fused latent features showed excellent performance in the class separability experiment,and the algorithm could be iterated to convergence with superior classification performance over other methods at the missing rates of 30%and 50%.Conclusion The proposed model has excellent performance in classification task of HGG and LGG and outperforms other non-holonomic multimodal classification models,demonstrating its potential for efficient processing of non-holonomic multimodal data.