Scholarlink

<p>This study aimed to identify tumor-associated autoantibodies (TAAbs) with diagnostic potential for the early detection of oral squamous cell carcinoma (OSCC). Bioinformatics analyses were used to screen candidate genes. The candidate tumor-associated antigens (TAAs) were selected from the proteins encoded by the candidate genes. Serum levels of corresponding TAAbs were measured by enzyme-linked immunosorbent assay (ELISA) in 496 participants. Eight machine learning algorithms were employed to develop diagnostic models, and Shapley Additive exPlanations (SHAP) were applied to interpret the optimal model. Twelve candidate genes were identified, among which eight encoded proteins were confirmed to be overexpressed in OSCC. Based on mRNA expression evidence, all 12 encoded proteins were included as candidate TAAs. Of the corresponding autoantibodies, five TAAbs (anti-BLM, anti-BUB1, anti-KIF18A, anti-KIF2C, and anti-TPX2) demonstrated potential diagnostic performance in both the training and validation sets. Among the eight models constructed, the Naive Bayes (NB) model performed best, achieving an area under the receiver operating characteristic curve (AUC) of 0.75 (95% CI 0.70–0.80) in the training set and 0.66 (95% CI 0.57–0.75) in the validation set. SHAP analysis indicated anti-KIF2C contributed most to predictive performance. Five TAAbs were identified with diagnostic potential for OSCC. The NB model constructed based on these TAAbs demonstrated potential diagnostic performance.</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

Early detection of oral squamous cell carcinoma using five tumor-associated autoantibodies and a Naive Bayes-based machine learning model

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月15日

查看全文
  • Lijuan Xu,
  • Weihong Xie,
  • Yuanlin Zou,
  • Qian Yang,
  • Zhong Zheng,
  • Xiaoyue Zhang,
  • Yihe Hou,
  • Yuqi Liu,
  • Meng Li,
  • Hua Ye,
  • Peng Wang

摘要

This study aimed to identify tumor-associated autoantibodies (TAAbs) with diagnostic potential for the early detection of oral squamous cell carcinoma (OSCC). Bioinformatics analyses were used to screen candidate genes. The candidate tumor-associated antigens (TAAs) were selected from the proteins encoded by the candidate genes. Serum levels of corresponding TAAbs were measured by enzyme-linked immunosorbent assay (ELISA) in 496 participants. Eight machine learning algorithms were employed to develop diagnostic models, and Shapley Additive exPlanations (SHAP) were applied to interpret the optimal model. Twelve candidate genes were identified, among which eight encoded proteins were confirmed to be overexpressed in OSCC. Based on mRNA expression evidence, all 12 encoded proteins were included as candidate TAAs. Of the corresponding autoantibodies, five TAAbs (anti-BLM, anti-BUB1, anti-KIF18A, anti-KIF2C, and anti-TPX2) demonstrated potential diagnostic performance in both the training and validation sets. Among the eight models constructed, the Naive Bayes (NB) model performed best, achieving an area under the receiver operating characteristic curve (AUC) of 0.75 (95% CI 0.70–0.80) in the training set and 0.66 (95% CI 0.57–0.75) in the validation set. SHAP analysis indicated anti-KIF2C contributed most to predictive performance. Five TAAbs were identified with diagnostic potential for OSCC. The NB model constructed based on these TAAbs demonstrated potential diagnostic performance.