Scholarlink

Background: Multi-omic integration drives cancer subtyping, therapy outcome prediction, and personalised treatment. This review examines advanced integration approaches and their applications in oncology. Methods: We analysed 20 studies published between 2019 and 2024, categorised into early, intermediate and late integration approaches. Datasets include RNA-Seq, DNA methylation, miRNA-Seq, protein expression and CNV data. Results: Intermediate integration (75%) using autoencoders and graph networks achieved AUCs >0.95; early (15%) is simpler but limited by heterogeneity; late (10%) enabled NSCLC diagnosis (F1: 96.8%, AUC: 0.993), with challenges in heterogeneity, computation, and standardisation. Conclusion: Intermediate integration ensures robustness; late integration suits multi-scale data; addressing challenges is key to clinical adoption.

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

Comparative Analysis of Multi-omic Integration Approaches for Tumor Subtype Classification: A Systematic Literature Review

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Fatima Zahra Belharar,
  • Ouafae Belharar,
  • Sara Retal,
  • Nassim Kharmoum,
  • Soumia Ziti

摘要

Background: Multi-omic integration drives cancer subtyping, therapy outcome prediction, and personalised treatment. This review examines advanced integration approaches and their applications in oncology. Methods: We analysed 20 studies published between 2019 and 2024, categorised into early, intermediate and late integration approaches. Datasets include RNA-Seq, DNA methylation, miRNA-Seq, protein expression and CNV data. Results: Intermediate integration (75%) using autoencoders and graph networks achieved AUCs >0.95; early (15%) is simpler but limited by heterogeneity; late (10%) enabled NSCLC diagnosis (F1: 96.8%, AUC: 0.993), with challenges in heterogeneity, computation, and standardisation. Conclusion: Intermediate integration ensures robustness; late integration suits multi-scale data; addressing challenges is key to clinical adoption.