AI-Driven Multi-Omics Integrated Applications Using Diverse Neural Networks for Breast Cancer Diagnostic Screening and Biomarker Discovery.

Abstract

Breast cancer (BC) is a highly complex and heterogeneous malignancy and the most prevalent cancer among women worldwide. The diagnosis, prognosis, and the treatment of BC pose significant challenges that are responsible for their limited therapeutic efficacy. Omics-based technologies have gained substantial attention in BC diagnosis through molecular profiling and diverse clinical analytics. The integration of metabolomics, proteomics, transcriptomics, and genomics provides a multidimensional approach to personalized BC diagnosis and treatment through high-throughput molecular profiling. Moreover, the emergence of artificial intelligence (AI) has also supported more accurate and early diagnosis of BC through multimodal integration of diverse datasets. The integration of advanced deep learning (DL) and machine learning (ML) has been extensively exploited for tumor grading, histopathological classification, molecular profiling, diagnostic imaging, and prognostic prediction. This review aims to summarize recent developments in AI-driven multi-omics approaches for the discovery of BC biomarkers. We have also highlighted the integration of omics-based data like metabolomics, proteomics, transcriptomics, and genomics with key AI techniques, including ML and DL, that play a crucial role in the inclusion of multi-omics in cancer and biomarker discovery. We have further discussed AI-based BC screening and diagnostic approaches, as well as the contribution of AI models for patient stratification, biomarker discovery, and prediction of therapeutic response. Additionally, key limitations and challenges, including data heterogeneity, high computational complexity, and model interpretability, have also been highlighted in the present review. Conclusively, we have also outlined future perspectives on the integration of AI and multi-omics to revolutionize precision clinical medicine and improve clinical outcomes in BC theranostics.