Scholarlink

<p>Fulminant myocarditis (FM) is a rare but life-threatening pediatric condition that can rapidly progress to cardiogenic shock and fatal arrhythmia. Prognostic biomarkers in FM are essential for optimizing treatment strategies. Although inflammatory cytokines have been associated with the pathogenesis of FM, their prognostic value remains unclear. This study aimed to identify mortality-associated markers by integrating cytokine profiles and clinical variables through a machine learning approach. We retrospectively analyzed 21 pediatric FM cases from two tertiary centers (2012–2022). At admission, 37 cytokines and 14 clinical parameters were assessed. Partial least squares discriminant analysis was employed to identify prognostic features, with variable importance in projection scores quantifying their contribution. Model performance was evaluated using repeated stratified 3-fold cross-validation with 100 repeats. For each repeat, predictions obtained from the three folds were averaged at the sample level, and performance metrics were computed from these fold-averaged predictions. Statistical significance was determined via the Benjamini–Hochberg method at a false discovery rate of 0.05. Of the 51 features analyzed, 18 emerged as key predictors, 15 cytokines and 3 clinical parameters, with variable importance in projection (VIP) scores above 1.0. Seven cytokines (TNF-α, MIP-1α, M-CSF, IL-8, IL-6, IL-15, and IP-10) were both statistically significant and highly important. TNF-α had the highest VIP score among all predictors. Three clinical parameters (CK-MB, pH, and lactate) were also linked to poor outcomes. The model performed robustly, with an area under the receiver operating characteristic curve (AUROC) of 0.912, an area under the precision–recall curve (AUPRC) of 0.874, 83.8% accuracy, 71.0% sensitivity, and 90.1% specificity. Furthermore, the cytokine-enriched model showed higher predictive performance than the clinical-only model across accuracy, AUROC, and AUPRC. Analysis of cytokine profiles using machine learning may identify biomarkers associated with mortality risk in pediatric FM. TNF-α emerged as a key cytokine associated with mortality, supporting its potential role as a prognostic biomarker. Importantly, cytokine markers added prognostic value beyond routine clinical variables.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine learning-based mortality prediction for pediatric fulminant myocarditis using cytokine profiles

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月24日

查看全文
  • Sihuan Jing,
  • Takanori Suzuki,
  • Kenta T. Suzuki,
  • Yoji Nomura,
  • Katsuyuki Kunida,
  • Yuichi Sakumura,
  • Hidetoshi Uchida,
  • Kazuyoshi Saito,
  • Ryoichi Ito,
  • Machiko Kito,
  • Satoru Kawai,
  • Alejandro A. Floh,
  • Aamir Jeewa,
  • Junichiro Yoshimoto,
  • Tetsushi Yoshikawa,
  • Kazushi Yasuda

摘要

Fulminant myocarditis (FM) is a rare but life-threatening pediatric condition that can rapidly progress to cardiogenic shock and fatal arrhythmia. Prognostic biomarkers in FM are essential for optimizing treatment strategies. Although inflammatory cytokines have been associated with the pathogenesis of FM, their prognostic value remains unclear. This study aimed to identify mortality-associated markers by integrating cytokine profiles and clinical variables through a machine learning approach. We retrospectively analyzed 21 pediatric FM cases from two tertiary centers (2012–2022). At admission, 37 cytokines and 14 clinical parameters were assessed. Partial least squares discriminant analysis was employed to identify prognostic features, with variable importance in projection scores quantifying their contribution. Model performance was evaluated using repeated stratified 3-fold cross-validation with 100 repeats. For each repeat, predictions obtained from the three folds were averaged at the sample level, and performance metrics were computed from these fold-averaged predictions. Statistical significance was determined via the Benjamini–Hochberg method at a false discovery rate of 0.05. Of the 51 features analyzed, 18 emerged as key predictors, 15 cytokines and 3 clinical parameters, with variable importance in projection (VIP) scores above 1.0. Seven cytokines (TNF-α, MIP-1α, M-CSF, IL-8, IL-6, IL-15, and IP-10) were both statistically significant and highly important. TNF-α had the highest VIP score among all predictors. Three clinical parameters (CK-MB, pH, and lactate) were also linked to poor outcomes. The model performed robustly, with an area under the receiver operating characteristic curve (AUROC) of 0.912, an area under the precision–recall curve (AUPRC) of 0.874, 83.8% accuracy, 71.0% sensitivity, and 90.1% specificity. Furthermore, the cytokine-enriched model showed higher predictive performance than the clinical-only model across accuracy, AUROC, and AUPRC. Analysis of cytokine profiles using machine learning may identify biomarkers associated with mortality risk in pediatric FM. TNF-α emerged as a key cytokine associated with mortality, supporting its potential role as a prognostic biomarker. Importantly, cytokine markers added prognostic value beyond routine clinical variables.