Artificial Intelligence and Machine Learning Techniques Using Omics Data for Cancer Diagnosis and Treatment

Abstract

AbstractCancer is a heterogeneous disease concerning molecular, functional and clinical behaviour, and poses a challenge for timely detection and treatment. Early detection and prognosis of cancer type may facilitate refined clinical management of cancer treatment. Recent technological development, such as next-generation sequencing, generated a large number of omics datasets in cancer genomics. The genome-wide biological information, such as cancer driver mutations, aberrantly methylated regions, gene, and miRNA expression profiles, is helpful for predicting the cancer onset, subtypes, and treatment response and is valuable for improving diagnosis and therapeutic and clinical decisions. In this context, machine learning (ML) algorithms and artificial intelligence have been beneficial and essential for the better accuracy of cancer-related predictions. Here, we mainly focus on research based on these omics data, paying close attention to machine learning methods. We summarize various kinds of omics data and different ML algorithms effective in cancer prediction. We also highlighted the applications of the ML algorithm on genomic information in cancer, including cancer classification, therapy response, survival, metastasis, and biomarker identification. Further we discussed the novel approaches in machine learning for improving cancer prediction. These data-driven approaches can potentially provide a new solution for enhancing the precise treatment of cancer.