Objective:To develop and validate machine learning-based radiomics models using preoperative CT images for individualized prediction of mitotic index (MI) in patients with gastrointestinal stromal tumors (GIST).Methods:The study was a case-control study. The data of 348 GIST patients confirmed by pathology were retrospectively collected from two independent medical centers: the Affiliated Hospital of Qingdao University (center 1) and Shandong Provincial Hospital Affiliated to Shandong First Medical University (center 2), covering the period from January 2013 to June 2018. Patients from center 1 were divided into a training cohort (176 cases) and an internal validation cohort (75 cases) at a ratio of 7∶3 using random sampling. Patients from center 2 served as an independent external validation cohort (97 cases). The primary endpoint was MI, categorized into high MI (145 cases) and low MI (203 cases) groups. Radiomic features were extracted from the portal venous phase images of preoperative contrast-enhanced CT scans. Five machine learning algorithms, including logistic regression, support vector machine, random forest, decision tree, and extreme gradient boosting (XGBoost),were employed to construct MI prediction models. The optimal model was identified using receiver operating characteristic curves. An individualized prediction model was developed by integrating the the optimal machine learning model combined with selected independent clinical factors, and the importance of features was visualized using Shapley Additive Explanation (SHAP) analysis. Patients were followed up, and Kaplan-Meier curves along with log-rank tests were used to evaluate recurrence-free survival (RFS) differences between the predicted high MI and low MI groups.Results:Among the five constructed machine learning models, the XGBoost model demonstrated the best predictive performance, with area under the curve (AUC) of 0.809 (95% CI 0.738-0.872), 0.693 (95% CI 0.571-0.809), and 0.718 (95% CI 0.605-0.822) in the training cohort, internal validation cohort, and external validation cohort, respectively. An individualized prediction model combining the XGBoost model with independent clinical factors (tumor location and tumor size) was developed. The model achieved AUC of 0.843 (95% CI 0.785-0.899), 0.791 (95% CI 0.680-0.894), and 0.777 (95% CI 0.678-0.861) in the training cohort, internal validation cohort, and external validation cohort, respectively. SHAP analysis indicated that radiomic features had the highest predictive impact. In both the training cohort and internal validation cohort, the RFS of patients predicted to be in the high MI group was lower than that of the low MI group, with statistically significant differences ( χ2=14.58, 9.52, both P<0.001). However, there was no statistically significant difference in RFS in the external validation set ( χ2=6.18, P=0.080). Conclusions:The optimal XGBoost model based on radiomic features extracted from preoperative portal venous phase CT images, when combined with clinical factors, can effectively predict the MI of GIST patients.

Object:To explore the feasibility of using artificial intelligence for quality control of echocardiographic images.Methods:Retrospectively，5 000 two-dimensional echocardiographic video images within the period from 2021 to 2023 were randomly retrieved from the echocardiography database of Nanjing Drum Tower Hospital，Affiliated Hospital of Medical School，Nanjing University. Among these selected images，1 559 of them were apical views. The physician team formulated the scoring rules，which specifically included four scoring criteria：gain，scaling ratio，cardiac axis angle，and structure. Subsequently，the data were labeled with view classification and image quality scores. The labeled data were further partitioned into the training set（ n = 643），the validation set（ n = 276），and the test set（ n = 640）. The training and validation sets were utilized for constructing the models for view classification and quality assessment，while the test set was employed to verify the models' effectiveness. The view classification module was implemented using the SlowFast model，and the quality assessment module involved algorithms such as ResNet，Video Swin Transformer，SSD，and U-Net. Results:The average accuracy，precision，recall rate and F1 score of the classification model in identifying each apical view were 0.987 1，0.983 0，0.987 1 and 0.984 9 respectively，and the inference time was（333.4 ± 105.4）ms. The average accuracies of the quality assessment module in terms of gain，scaling ratio，cardiac axis angle and display of main structures were 0.915 1，0.928 2，0.938 7 and 0.965 6 respectively，and the overall scoring accuracy was 0.912 7.Conclusions:The echocardiogram quality control system developed in this research can effectively classify and evaluate the quality of two-dimensional images of the apical views in echocardiograms. Moreover，it guarantees the objectivity，timeliness and high-efficiency of quality control，which has reference value for the establishment of the echocardiogram quality control system.

Genomic imprinting refers to the phenomenon of differential expression of two alleles due to their different parental origins. Genes that produce genomic imprinting are usually called imprinted genes. The genetic effect caused by the presence of imprinted genes is called parent-of-origin effect. Parent-of-origin effect and genomic imprinting play important roles in the pathophysiological mechanism and occurrence and development of cardio-metabolic diseases. In-depth exploration of the law and potential roles of imprinted genes and parent-of-origin effects will help to better understand the mechanism of cardio-metabolic diseases, and also provide important theoretical basis for the precise treatment of diseases related to imprinted genes.

Objective:To explore the spousal correlations of total cholesterol(TC),total triglyceride(TG),low-density lipoprotein cholesterol(LDL-C),and high-density lipoprotein cholesterol(HDL-C),and to investigate the reasons behind these spousal correlations.Methods:Participants and data were from the baseline survey of family-based cohort studies in Fangshan,Beijing and Tulou,Fujian.The ori-gin of spousal correlations were explored from perspectives of convergence,assortative mating,social ho-mogamy.Pearson's correlation and generalized linear models(GLM)were used to estimate the spousal correlation.Convergence was assessed by Pearson's correlation between the phenotypic differences be-tween couples and the duration of marriage,with GLM used for further validation.Pearson's correlation of genetic risk scores(GRS)and couple-specific Mendelian randomization(MR)were calculated to assess the genetic correlation and possible causal relationships between spouses.Two-independent-sample t-tests were used to compare GRS consistency across subgroups divided by education attainment,couple-specific MR and Q statistics used to test assortative mating in subgroups and intergroup differences.Results:In the study,342 couples(287 couples from Fangshan and 55 couples from Fujian)were included,with the average age of(64.91±8.76)years.Spousal correlations of TC,TG,HDL-C,and LDL-C showed statistically significant associations both before and after adjusting for covariates,with effect sizes of 0.229(95％CI:0.125-0.327),0.257(95％CI:0.155-0.354),0.179(95％CI:0.074-0.280),and 0.181(95％CI:0.076-0.282).For convergence,for each additional year of marriage,ΔTC increased by 0.016 mmol/L(95％CI:0.001-0.033 mmol/L),and ΔLDL-C increased by 0.017 mmol/L(95％CI:0.002-0.031 mmol/L).For assortative mating,GRS correlations and results of couple specific MR didn't show any statistical significance.For social homogamy,no differences in GRS or assortative mating were found between subgroups stratified by education attainment.Conclusion:The blood lipid in participants exhibit spousal phenotypic correlations,however,no effects of convergence,assortative mating or social homogamy were observed.More independent studies with larger sample sizes are warranted to further validate these findings in the future.

Genomic imprinting refers to the phenomenon of differential expression of two alleles due to their different parental origins. Genes that produce genomic imprinting are usually called imprinted genes. The genetic effect caused by the presence of imprinted genes is called parent-of-origin effect. Parent-of-origin effect and genomic imprinting play important roles in the pathophysiological mechanism and occurrence and development of cardio-metabolic diseases. In-depth exploration of the law and potential roles of imprinted genes and parent-of-origin effects will help to better understand the mechanism of cardio-metabolic diseases, and also provide important theoretical basis for the precise treatment of diseases related to imprinted genes.

Objective:To explore the spousal correlations of total cholesterol(TC),total triglyceride(TG),low-density lipoprotein cholesterol(LDL-C),and high-density lipoprotein cholesterol(HDL-C),and to investigate the reasons behind these spousal correlations.Methods:Participants and data were from the baseline survey of family-based cohort studies in Fangshan,Beijing and Tulou,Fujian.The ori-gin of spousal correlations were explored from perspectives of convergence,assortative mating,social ho-mogamy.Pearson's correlation and generalized linear models(GLM)were used to estimate the spousal correlation.Convergence was assessed by Pearson's correlation between the phenotypic differences be-tween couples and the duration of marriage,with GLM used for further validation.Pearson's correlation of genetic risk scores(GRS)and couple-specific Mendelian randomization(MR)were calculated to assess the genetic correlation and possible causal relationships between spouses.Two-independent-sample t-tests were used to compare GRS consistency across subgroups divided by education attainment,couple-specific MR and Q statistics used to test assortative mating in subgroups and intergroup differences.Results:In the study,342 couples(287 couples from Fangshan and 55 couples from Fujian)were included,with the average age of(64.91±8.76)years.Spousal correlations of TC,TG,HDL-C,and LDL-C showed statistically significant associations both before and after adjusting for covariates,with effect sizes of 0.229(95％CI:0.125-0.327),0.257(95％CI:0.155-0.354),0.179(95％CI:0.074-0.280),and 0.181(95％CI:0.076-0.282).For convergence,for each additional year of marriage,ΔTC increased by 0.016 mmol/L(95％CI:0.001-0.033 mmol/L),and ΔLDL-C increased by 0.017 mmol/L(95％CI:0.002-0.031 mmol/L).For assortative mating,GRS correlations and results of couple specific MR didn't show any statistical significance.For social homogamy,no differences in GRS or assortative mating were found between subgroups stratified by education attainment.Conclusion:The blood lipid in participants exhibit spousal phenotypic correlations,however,no effects of convergence,assortative mating or social homogamy were observed.More independent studies with larger sample sizes are warranted to further validate these findings in the future.

Objective:To develop and validate machine learning-based radiomics models using preoperative CT images for individualized prediction of mitotic index (MI) in patients with gastrointestinal stromal tumors (GIST).Methods:The study was a case-control study. The data of 348 GIST patients confirmed by pathology were retrospectively collected from two independent medical centers: the Affiliated Hospital of Qingdao University (center 1) and Shandong Provincial Hospital Affiliated to Shandong First Medical University (center 2), covering the period from January 2013 to June 2018. Patients from center 1 were divided into a training cohort (176 cases) and an internal validation cohort (75 cases) at a ratio of 7∶3 using random sampling. Patients from center 2 served as an independent external validation cohort (97 cases). The primary endpoint was MI, categorized into high MI (145 cases) and low MI (203 cases) groups. Radiomic features were extracted from the portal venous phase images of preoperative contrast-enhanced CT scans. Five machine learning algorithms, including logistic regression, support vector machine, random forest, decision tree, and extreme gradient boosting (XGBoost),were employed to construct MI prediction models. The optimal model was identified using receiver operating characteristic curves. An individualized prediction model was developed by integrating the the optimal machine learning model combined with selected independent clinical factors, and the importance of features was visualized using Shapley Additive Explanation (SHAP) analysis. Patients were followed up, and Kaplan-Meier curves along with log-rank tests were used to evaluate recurrence-free survival (RFS) differences between the predicted high MI and low MI groups.Results:Among the five constructed machine learning models, the XGBoost model demonstrated the best predictive performance, with area under the curve (AUC) of 0.809 (95% CI 0.738-0.872), 0.693 (95% CI 0.571-0.809), and 0.718 (95% CI 0.605-0.822) in the training cohort, internal validation cohort, and external validation cohort, respectively. An individualized prediction model combining the XGBoost model with independent clinical factors (tumor location and tumor size) was developed. The model achieved AUC of 0.843 (95% CI 0.785-0.899), 0.791 (95% CI 0.680-0.894), and 0.777 (95% CI 0.678-0.861) in the training cohort, internal validation cohort, and external validation cohort, respectively. SHAP analysis indicated that radiomic features had the highest predictive impact. In both the training cohort and internal validation cohort, the RFS of patients predicted to be in the high MI group was lower than that of the low MI group, with statistically significant differences ( χ2=14.58, 9.52, both P<0.001). However, there was no statistically significant difference in RFS in the external validation set ( χ2=6.18, P=0.080). Conclusions:The optimal XGBoost model based on radiomic features extracted from preoperative portal venous phase CT images, when combined with clinical factors, can effectively predict the MI of GIST patients.

Object:To explore the feasibility of using artificial intelligence for quality control of echocardiographic images.Methods:Retrospectively，5 000 two-dimensional echocardiographic video images within the period from 2021 to 2023 were randomly retrieved from the echocardiography database of Nanjing Drum Tower Hospital，Affiliated Hospital of Medical School，Nanjing University. Among these selected images，1 559 of them were apical views. The physician team formulated the scoring rules，which specifically included four scoring criteria：gain，scaling ratio，cardiac axis angle，and structure. Subsequently，the data were labeled with view classification and image quality scores. The labeled data were further partitioned into the training set（ n = 643），the validation set（ n = 276），and the test set（ n = 640）. The training and validation sets were utilized for constructing the models for view classification and quality assessment，while the test set was employed to verify the models' effectiveness. The view classification module was implemented using the SlowFast model，and the quality assessment module involved algorithms such as ResNet，Video Swin Transformer，SSD，and U-Net. Results:The average accuracy，precision，recall rate and F1 score of the classification model in identifying each apical view were 0.987 1，0.983 0，0.987 1 and 0.984 9 respectively，and the inference time was（333.4 ± 105.4）ms. The average accuracies of the quality assessment module in terms of gain，scaling ratio，cardiac axis angle and display of main structures were 0.915 1，0.928 2，0.938 7 and 0.965 6 respectively，and the overall scoring accuracy was 0.912 7.Conclusions:The echocardiogram quality control system developed in this research can effectively classify and evaluate the quality of two-dimensional images of the apical views in echocardiograms. Moreover，it guarantees the objectivity，timeliness and high-efficiency of quality control，which has reference value for the establishment of the echocardiogram quality control system.

BACKGROUND:There is an increasing demand for orthodontic treatment,and periodontally accelerated osteogenic orthodontics(PAOO)technique can make it possible to move orthodontic teeth that are limited by thin alveolar bone. OBJECTIVE:To investigate the biomechanics of orthodontic tooth movement before and after periodontally accelerated osteogenic orthodontics(PAOO)surgery to increase alveolar bone volume using the three-dimensional finite element method. METHODS:A patient undergoing PAOO surgery before orthodontic treatment to increase bone volume on the labial side of the mandibular anterior region was selected.The patient was under invisible orthodontics.Two three-dimensional finite element models were constructed based on the patient's preoperative and 6-month postoperative cone beam CT data.Both models simulated the movement of tooth 33:experiment Ⅰ:distal-central movement of 0.25 mm;experiment Ⅱ:lingual movement of 0.25 mm;and experiment Ⅲ:intrusion movement of 0.10 mm.The stress distribution and initial displacement trend of tooth 33,periodontal ligament and surrounding alveolar bone under the action of the invisible aligner were analyzed before and after the PAOO procedure. RESULTS AND CONCLUSION:Dental stress analysis:In the same orthodontic tooth movement,the maximum Von-Mises stress and overall stress values of tooth 33 were all higher before surgery than after surgery;there were similar distribution areas of maximum equivalent stress and overall distribution trends of Von-Mises stress before and after surgery.Periodontal ligament stress analysis:In the same orthodontic tooth movement,the maximum Von-Mises stress and overall stress values of the periodontal ligament were higher before surgery than after surgery,and there were similar distribution areas of the maximum equivalent stress and overall distribution trends of Von-Mises stress before and after surgery.Alveolar bone stress analysis:In the same orthodontic tooth movement,the maximum Von-Mises stress values of the alveolar bone around tooth 33 were higher before surgery than after surgery,while the equivalent stress distribution showed a gradual decrease from the top of the alveolar ridge to the root.Initial displacement analysis:In the same orthodontic tooth movement,the initial displacements in the main displacement direction for all six observation points of tooth 33 were smaller before surgery than after surgery,and showed a tendency to gradually decrease from the tooth tip to the apex.Therefore,there were differences in the biomechanical characteristics of orthodontic tooth movement before and after the PAOO surgery.With the clear aligner,the postoperative equivalent stress values on the dentition,periodontal ligament,and surrounding alveolar bone were lower than before the surgery,and the initial displacements of the orthodontic teeth after the surgery are larger than before.These findings suggest that PAOO can release the restriction of thin alveolar bone on the movement of orthodontic tooth by increasing alveolar bone thickness,effectively improving the force on the roots,periodontal ligament,and alveolar bone,avoiding the stress concentration on orthodontic tooth in the thin alveolar bone area that can cause complications when moving,and improving the efficiency of tooth movement.

Objective:To explore the effects of short-term particulate matter(PM)exposure and the melatonin receptor 1B(MTNR1B)gene on triglyceride-glucose(TyG)index utilizing data from Fang-shan Family-based Ischemic Stroke Study in China(FISSIC).Methods:Probands and their relatives from 9 rural areas in Fangshan District,Beijing,were included in the study.PM data were obtained from fixed monitoring stations of the National Air Pollution Monitoring System.TyG index was calculated by fasting triglyceride and glucose concentrations.The associations of short-term PM exposure and rs10830963 polymorphism of the MTNR1B gene with the TyG index were assessed using mixed linear models,in which covariates such as age,sex,and lifestyles were adjusted for.Gene-environment inter-action analysis was furtherly performed using the maximum likelihood methods to explore the potential effect modifier role of rs10830963 polymorphism in the association of PM with TyG index.Results:A total of 4 395 participants from 2 084 families were included in the study,and the mean age of the study participants was(58.98±8.68)years,with 53.90％females.The results of association analyses showed that for every 10 μg/m3 increase in PM2.5 concentration,TyG index increased by 0.017(95％CI:0.007-0.027),while for per 10 μg/m3 increment in PM1o,TyG index increased by 0.010(95％CI:0.003-0.017).And the associations all had lagged effects.In addition,there was a positive association between the rs10830963 polymorphism and the TyG index.For per increase in risk allele G,TyG index was elevated by 0.040(95％CI:0.004-0.076).The TyG index was 0.079(95％CI:0.005-0.152)higher in carriers of the GG genotype compared with carriers of the CC genotype.The inter-action of rs10830963 polymorphism with PM exposure had not been found to be statistically significant in the present study.Conclusion:Short-term exposure to PM2.5 and PM10 were associated with higher TyG index.The G allele of rs10830963 polymorphism in the MTNR1B gene was associated with the elevated TyG index.