Modern society is rapidly turning to an aging society as the average lifespan has increased thanks to the development of medical technologies and improvement in living standards. The effective treatment of diseases is critical to not only the health of individual patients, but also to the society and economy. Efforts have long been made to develop new drugs for the effective treatment of diseases, but the number of newly approved drugs (~40 drugs per year) is still unsatisfactory, considering the input time and cost of new drug development as well as the number of rare diseases being greater than 7,800. One of the strategies to solve this problem is Multi-Component Multi-Target (MCMT) approach, where multiple drugs can target multiple target proteins at the same time. However, MCMT approach requires various platform technologies. The purpose of the present thesis study is to develop methods required for MCMT approach. In this regard, Chapter 1 analyzes druggability of the traditional oriental medicine (TOM) compounds by analyzing metabolite likeness. TOM compounds have a greater structural similarity with the metabolites in the human body than the approved drugs, and thus they may target multiple proteins. In addition, a large-scale literature analysis showed that the prescription principle of TOM is consistent with the mechanisms of drug synergism in modern medicine. Chapter 2 develops a drug interaction prediction method, DeepDDI based on structural similarity profile and deep neural network, that accurately predicts 45 types of drug-drug interactions for a given drug pair. DeepDDI was used to analyze not only drug-drug interactions, but also the interactions between all the approved drugs and 800 natural products. DeepDDI enabled the prediction of the drug interactions well as the prediction of activities of the natural products. Chapter 3 introduce a genome-scale metabolic model (GEM), which is a platform technology that predicts the metabolic phenotype at systems level. Chapter 4 and 5 suggest a strategy for developing accurate human GEM, and Chapter5 develops a computational framework, Gene-Transcript-Protein-Reaction Associations (GeTPRA), that systematically generates metabolic reactions that could be generated from the human alternative splicing, and adds the predicted metabolic reactions to the human GEM, Recon 2M.2. The resulting human GEM was demonstrated to accurately predict metabolic phenotypes. Chapter 6 introduces a strategy for analyzing metabolic characteristics at the system level using multi-omics data and human GEM integrated with transcriptional regulatory network. In particular, this strategy was used to analyze the metabolic characteristics of liver cancer stem cells. Through this analysis, major transcription factors involved in metabolic regulation of liver cancer stem cells were predicted and experimentally validated. Finally, Chapter 7 summarizes the significance of the results obtained by the present thesis study.