Validation of an established TW3 artificial intelligence bone age assessment system: a prospective, multicenter, confirmatory study.

Background: In 2020, our center established a Tanner-Whitehouse 3 (TW3) artificial intelligence (AI) system using a convolutional neural network (CNN), which was built upon 9059 radiographs. However, the system, upon which our study is based, lacked a gold standard for comparison and had not undergone thorough evaluation in different working environments. Methods: To further verify the applicability of the AI system in clinical bone age assessment (BAA) and to enhance the accuracy and homogeneity of BAA, a prospective multi-center validation was conducted. This study utilized 744 left-hand radiographs of patients, ranging from 1 to 20 years of age, with 378 boys and 366 girls. These radiographs were obtained from nine different children's hospitals between August and December 2020. The BAAs were performed using the TW3 AI system and were also reviewed by experienced reviewers. Bone age accuracy within 1 year, root mean square error (RMSE), and mean absolute error (MAE) were statistically calculated to evaluate the accuracy. Kappa test and Bland-Altman (B-A) plot were conducted to measure the diagnostic consistency. Results: The system exhibited a high level of performance, producing results that closely aligned with those of the reviewers. It achieved a RMSE of 0.52 years and an accuracy of 94.55% for the radius, ulna, and short bones series. When assessing the carpal series of bones, the system achieved a RMSE of 0.85 years and an accuracy of 80.38%. Overall, the system displayed satisfactory accuracy and RMSE, particularly in patients over 7 years old. The system excelled in evaluating the carpal bone age of patients aged 1-6. Both the Kappa test and B-A plot demonstrated substantial consistency between the system and the reviewers, although the model encountered challenges in consistently distinguishing specific bones, such as the capitate. Furthermore, the system's performance proved acceptable across different genders and age groups, as well as radiography instruments. Conclusions: In this multi-center validation, the system showcased its potential to enhance the efficiency and consistency of healthy delivery, ultimately resulting in improved patient outcomes and reduced healthcare costs.

Regional Disparities in Obesity Among a Heterogeneous Population of Chinese Children and Adolescents.

Importance: Obesity is a public health challenge in China, but the geographical profiles of overweight and obesity among Chinese children are limited. Objective: To examine regional disparities in the prevalence of obesity among the heterogeneous population of Chinese children and adolescents to provide a more accurate profile of obesity among children in China. Design, Setting, and Participants: The Prevalence and Risk Factors for Obesity and Diabetes in Youth (PRODY) study was a cross-sectional survey study conducted from January 1, 2017, to December 31, 2019, among 201 098 children aged 3 to 18 years from 11 provinces, autonomous regions, and municipalities that produced a sample of Chinese children with a full range of ages and wide geographical coverage using a multistage, stratified, cluster-sampling design. Exposures: Five regions geographically representative of China (northern, eastern, southern, western, and central). Main Outcomes and Measures: The body weights and heights of all participants were measured. Multilevel, multinomial logistic regression models were used to estimate the prevalence of overweight and obesity. Results: Among 201 098 healthy children (105 875 boys [52.6%]; mean [SD] age, 9.8 [3.8] years) from eastern, southern, northern, central, and western China, the highest obesity prevalence was estimated for children aged 8 to 13 years in northern China (from 18.8% [95% CI, 16.2%-21.7%] to 23.6% [95% CI, 20.5%-26.9%]) and for boys aged 3 to 6 years in western China (from 18.1% [95% CI, 10.4%-29.4%] to 28.6% [95% CI, 14.3%-49.0%]). Boys had a higher prevalence than girls of obesity only in eastern and northern China, with a mean difference in prevalence of 4.6% (95% CI, 3.8%-5.4%) and 7.6% (95% CI, 6.5%-8.6%), respectively. Conclusions and Relevance: In this survey study, substantial geographic disparities in the prevalence of obesity and overweight were found among the heterogeneous population of Chinese children. The results suggest that special attention should be paid to vulnerable children and that regionally adapted interventions are needed to efficiently mitigate obesity in children.

Diagnostic performance of convolutional neural network-based Tanner-Whitehouse 3 bone age assessment system.

BACKGROUND: Bone age can reflect the true growth and development status of a child; thus, it plays a critical role in evaluating growth and endocrine disorders. This study established and validated an optimized Tanner-Whitehouse 3 artificial intelligence (TW3-AI) bone age assessment (BAA) system based on a convolutional neural network (CNN). METHODS: A data set of 9,059 clinical radiographs of the left hand was obtained from the picture archives and communication systems (PACS) between January 2012 and December 2016. Among these, 8,005/9,059 (88%) samples were treated as the training set for model implementation, 804/9,059 (9%) samples as the validation set for parameters optimization, and the remaining 250/9,059 (3%) samples were used to verify the accuracy and reliability of the model compared to that of 4 experienced endocrinologists and 2 experienced radiologists. The overall variation of TW3-metacarpophalangeal, radius, ulna and short bones (RUS) and TW3-Carpal bone score, as well as each bone (13 RUS + 7 Carpal) between reviewers and the AI, were compared by Bland-Altman (BA) chart and Kappa test, respectively. Furthermore, the time consumption between the model and reviewers was also compared. RESULTS: The performance of TW3-AI model was highly consistent with the reviewers' overall estimation, and the root mean square (RMS) was 0.50 years. The accuracy of the BAA of the TW3-AI model was better than the estimate of the reviewers. Further analysis revealed that human interpretations of the male capitate, hamate, the first distal and fifth middle phalanx and female capitate, the trapezoid, and the third and fifth middle phalanx, were most inconsistent. The average image processing time was 1.5±0.2 s in the TW3-AI model, which was significantly shorter than manual interpretation. CONCLUSIONS: The diagnostic performance of CNN-based TW3 BAA was accurate and timesaving, and possesses better stability compared to diagnostics made by experienced experts.

TriGlycerides and high-density lipoprotein cholesterol ratio compared with homeostasis model assessment insulin resistance indexes in screening for metabolic syndrome in the chinese obese children: a cross section study.

BACKGROUND: Metabolic Syndrome (MS) is prevalant in China, especially according to the pediatric obesity group. Based on the MS-CHN2012 definition for Chinese children and adolescents the need to explore and establish a convienent MS screening become imminent. This study aims to investigate the optimal cut-off values, compare the accuracy for the (TriGlycerides (TG) to High-Density Lipoprotein Cholesterol (HDL-C)) (TG/HDL-C) ratio and Homeostasis Model Assessment Insulin Resistance (HOMA-IR) indexs to identify Metabolic Syndrome in obese pediatric population in China. METHOD: A total sample of 976 children (female 286 male 690, BMI > = 95 percentile) aged from 6-16 years underwent a medical assessment including a physical examination and investigations of total cholesterol, high-density lipoprotein, low-density lipoprotein, triglycerides, insulin, glucose, and oral glucose tolerance test to identify the components of Metabolic Syndrome. The validity and accuracy between TG/HDL-C ratio and HOMA-IR were compared by Receiver Operating Characteristics analysis (ROC). RESULT: TG/HDL-C ratio achieved a larger ROC Area under Curve (AUC = 0.843) than HOMA-IR indexes (0.640, 0.625 for HOMA1-IR, HOMA2-IR respectively) to screen for Metabolic Syndrome. The cut-off values for MS were: TG/HDL-C ratio > 1.25 (sensitivity: 80%; specificity: 75%), HOMA1-IR > 4.59 (sensitivity: 58.7%; specificity: 65.5%) and HOMA2-IR > 2.76 (sensitivity: 53.2%; specificity: 69.5%). The results kept robust after stratified by gender, age group and pubertal stage. DISCUSSION: TG/HDL-C ratio was a better indicator than the HOMA-IR to screen for a positive diagnosis for MS. Furthermore, the TG/HDL-C ratio was superior to the HOMA-IR indexes even after the control of possible confusions from the gender, age group and puberty stage. CONCLUSION: TG/HDL-C ratio proved a better index than HOMA-IR in screening for MS in obese children and adolescents. TG/HDL-C ratio has a discriminatory power in detecting potential MS in the Chinese obese pediatric population.

[Quantitative assessment of intrahepatic fat content in children and adolescents with non-alcoholic fatty liver disease].

OBJECTIVE: To quantitatively evaluate clinical significance of intrahepatic fat (IHF) content in children and adolescents with non-alcoholic fatty liver disease (NAFLD). METHODS: Ninety-three obese children were enrolled in this study. Physical parameters, liver function, serum lipids, glycemic and insulin related parameters were measured. Liver B-mode ultrasound (US) examination was performed. IHF content was quantified by 1H magnetic resonance spectroscopy (1H MRS). Three subgroups were classified according to the conditional diagnostic criteria for obese children: simple obesity (n=31), NAFLD-1 (US fatty liver and normal alanine aminotransterase, n=33) and NAFLD-2 (US fatty liver and elevated alanine aminotransterase, n=29). Twenty healthy age- and sex-matched children served as a control group. IHF content among the four groups was compared. The relationship of IHF content with other common clinical laboratory parameters and independent factors influencing increased IHF content were investigated. RESULTS: IHF content measured by 1H MRS was 0.80% (0.4%-1.0%), 2.9% (1.7%-4.30%), 14.0% (7.2%-17.5%) and 18.8% (14.0%-29.1%) respectively in the control, simple obese, NAFLD-1 and NAFLD-2 groups. There were significant differences in IHF content between the groups. Univariate correlation analysis demonstrated that IHF content was positively correlated with waist circumference, hip circumference, waisttohip ratio, body mass index, systolic blood pressure, diastolic blood pressure, alanine aminotransferase, aspartate aminoreansferase, γ-glutamic acid transtetase, triglyceride, low-density lipoprotein, OGTT 2-hour plasma glucose, fasting insulin, 2-hour insulin and insulin resisfence, and negatively correlated with high-density lipoprotein. Multivariate linear regression analysis demonstrated three independent risk factors for increased IHF content: increased waist circumference, increased 2-hour plasma glucose and decreased high-density lipoprotein levels. CONCLUSIONS: IHF content determined by 1H MRS can reflect early hepatic fatty infiltration and is closely related to the occurrence and progress of NAFLD in obese children and adolescents. There is a significant correlation between most of common clinical laboratory parameters and IHF content, and waist circumference, high-density lipoprotein and OGTT 2-hour plasma glucose are independent factors impacting IHF content.