Shunshun Cao, Aolei Chen, Botian Song, Yangyang Hu
{"title":"探讨甘油三酯-葡萄糖指数对青春期前儿童骨代谢的影响,回顾性研究:来自传统方法和基于机器学习的骨重塑预测的见解。","authors":"Shunshun Cao, Aolei Chen, Botian Song, Yangyang Hu","doi":"10.7717/peerj.19483","DOIUrl":null,"url":null,"abstract":"<p><strong>Background: </strong>Childhood obesity poses a significant risk to bone health, but the impact of insulin resistance (IR) on bone metabolism in prepubertal children, as assessed by the triglyceride-glucose (TyG) index, remains underexplored. Bone turnover markers (BTMs) provide a non-invasive method for evaluating bone remodeling, but their relationship to obesity-related metabolic changes requires further study.</p><p><strong>Methods: </strong>In this retrospective study of 332 prepubertal children (163 boys and 169 girls), we used multivariate linear regression and five machine learning (ML) algorithms to explore the association between the TyG index and BTMs, including β-C-terminal telopeptide of type 1 collagen (β-CTx), total procollagen type 1 N-terminal propeptide (T-P1NP), and N-terminal mid-fragment of osteocalcin (N-MID). The categorical boosting (CatBoost) models selected based on optimal performance metrics were interpreted using SHapley Additive exPlanation (SHAP) analysis to identify key features affecting prediction.</p><p><strong>Results: </strong>The TyG index was negatively correlated with β-CTx, T-P1NP, and N-MID levels (<i>P</i> < 0.05), with a dose-response effect. The CatBoost model showed higher predictive accuracy and robustness, with the area under the receiver operating characteristic curve (AUROC) values of 0.782 (95% CI [0.68-0.885]), 0.789 (95% CI [0.691-0.874]), and 0.727 (95% CI [0.619-0.827]) for β-CTx, T-P1NP, and N-MID predictions, respectively. The SHAP analysis highlighted body mass index (BMI) and HbA1c as the key predictors.</p><p><strong>Conclusions: </strong>The TyG index is a reliable predictor of bone metabolic disorders in prepubertal obese children, and the interpretable CatBoost model provides a cost-effective tool for early intervention. This study has important implications for prevention strategies for disorders of bone metabolism in prepubertal obese children to reduce the risk of skeletal fragility in adulthood or old age.</p>","PeriodicalId":19799,"journal":{"name":"PeerJ","volume":"13 ","pages":"e19483"},"PeriodicalIF":2.3000,"publicationDate":"2025-05-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12101447/pdf/","citationCount":"0","resultStr":"{\"title\":\"Exploring the effect of the triglyceride-glucose index on bone metabolism in prepubertal children, a retrospective study: insights from traditional methods and machine-learning-based bone remodeling prediction.\",\"authors\":\"Shunshun Cao, Aolei Chen, Botian Song, Yangyang Hu\",\"doi\":\"10.7717/peerj.19483\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p><strong>Background: </strong>Childhood obesity poses a significant risk to bone health, but the impact of insulin resistance (IR) on bone metabolism in prepubertal children, as assessed by the triglyceride-glucose (TyG) index, remains underexplored. Bone turnover markers (BTMs) provide a non-invasive method for evaluating bone remodeling, but their relationship to obesity-related metabolic changes requires further study.</p><p><strong>Methods: </strong>In this retrospective study of 332 prepubertal children (163 boys and 169 girls), we used multivariate linear regression and five machine learning (ML) algorithms to explore the association between the TyG index and BTMs, including β-C-terminal telopeptide of type 1 collagen (β-CTx), total procollagen type 1 N-terminal propeptide (T-P1NP), and N-terminal mid-fragment of osteocalcin (N-MID). The categorical boosting (CatBoost) models selected based on optimal performance metrics were interpreted using SHapley Additive exPlanation (SHAP) analysis to identify key features affecting prediction.</p><p><strong>Results: </strong>The TyG index was negatively correlated with β-CTx, T-P1NP, and N-MID levels (<i>P</i> < 0.05), with a dose-response effect. The CatBoost model showed higher predictive accuracy and robustness, with the area under the receiver operating characteristic curve (AUROC) values of 0.782 (95% CI [0.68-0.885]), 0.789 (95% CI [0.691-0.874]), and 0.727 (95% CI [0.619-0.827]) for β-CTx, T-P1NP, and N-MID predictions, respectively. The SHAP analysis highlighted body mass index (BMI) and HbA1c as the key predictors.</p><p><strong>Conclusions: </strong>The TyG index is a reliable predictor of bone metabolic disorders in prepubertal obese children, and the interpretable CatBoost model provides a cost-effective tool for early intervention. This study has important implications for prevention strategies for disorders of bone metabolism in prepubertal obese children to reduce the risk of skeletal fragility in adulthood or old age.</p>\",\"PeriodicalId\":19799,\"journal\":{\"name\":\"PeerJ\",\"volume\":\"13 \",\"pages\":\"e19483\"},\"PeriodicalIF\":2.3000,\"publicationDate\":\"2025-05-20\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12101447/pdf/\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"PeerJ\",\"FirstCategoryId\":\"99\",\"ListUrlMain\":\"https://doi.org/10.7717/peerj.19483\",\"RegionNum\":3,\"RegionCategory\":\"生物学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"2025/1/1 0:00:00\",\"PubModel\":\"eCollection\",\"JCR\":\"Q2\",\"JCRName\":\"MULTIDISCIPLINARY SCIENCES\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"PeerJ","FirstCategoryId":"99","ListUrlMain":"https://doi.org/10.7717/peerj.19483","RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"2025/1/1 0:00:00","PubModel":"eCollection","JCR":"Q2","JCRName":"MULTIDISCIPLINARY SCIENCES","Score":null,"Total":0}
Exploring the effect of the triglyceride-glucose index on bone metabolism in prepubertal children, a retrospective study: insights from traditional methods and machine-learning-based bone remodeling prediction.
Background: Childhood obesity poses a significant risk to bone health, but the impact of insulin resistance (IR) on bone metabolism in prepubertal children, as assessed by the triglyceride-glucose (TyG) index, remains underexplored. Bone turnover markers (BTMs) provide a non-invasive method for evaluating bone remodeling, but their relationship to obesity-related metabolic changes requires further study.
Methods: In this retrospective study of 332 prepubertal children (163 boys and 169 girls), we used multivariate linear regression and five machine learning (ML) algorithms to explore the association between the TyG index and BTMs, including β-C-terminal telopeptide of type 1 collagen (β-CTx), total procollagen type 1 N-terminal propeptide (T-P1NP), and N-terminal mid-fragment of osteocalcin (N-MID). The categorical boosting (CatBoost) models selected based on optimal performance metrics were interpreted using SHapley Additive exPlanation (SHAP) analysis to identify key features affecting prediction.
Results: The TyG index was negatively correlated with β-CTx, T-P1NP, and N-MID levels (P < 0.05), with a dose-response effect. The CatBoost model showed higher predictive accuracy and robustness, with the area under the receiver operating characteristic curve (AUROC) values of 0.782 (95% CI [0.68-0.885]), 0.789 (95% CI [0.691-0.874]), and 0.727 (95% CI [0.619-0.827]) for β-CTx, T-P1NP, and N-MID predictions, respectively. The SHAP analysis highlighted body mass index (BMI) and HbA1c as the key predictors.
Conclusions: The TyG index is a reliable predictor of bone metabolic disorders in prepubertal obese children, and the interpretable CatBoost model provides a cost-effective tool for early intervention. This study has important implications for prevention strategies for disorders of bone metabolism in prepubertal obese children to reduce the risk of skeletal fragility in adulthood or old age.
期刊介绍:
PeerJ is an open access peer-reviewed scientific journal covering research in the biological and medical sciences. At PeerJ, authors take out a lifetime publication plan (for as little as $99) which allows them to publish articles in the journal for free, forever. PeerJ has 5 Nobel Prize Winners on the Board; they have won several industry and media awards; and they are widely recognized as being one of the most interesting recent developments in academic publishing.