The field of deep learning-based drug-response prediction is examined in this overview of the literature, with an emphasis on more recent developments including multi-omics data. Accurate and customized medication response prediction techniques are becoming more and more important as precision medicine gains traction. Conventional methods frequently fail to capture the intricate relationships between biological variables affecting the effectiveness of drugs. The article looks at the difficulties that current approaches are facing and emphasizes how deep learning techniques can help to overcome these constraints. In particular, it addresses how to predict medication responses by integrating transcriptomics, proteomics, metabolomics, and genomes as a multi-omics data. Through the combined investigation of many chemical characteristics with deep neural networks, these techniques provide a thorough grasp of the biological principles that underpin therapeutic efficacy. The review summarizes new research findings and offers insights regarding multi-omics data.