Artificial Intelligence Enabled Drug Repurposing for Precision Therapeutics: A Systems Pharmacology Approach

Abstract

The pharmaceutical sector is caught in a critical situation, where the conventional de novo drug discovery has become unsustainable due to high costs, long development time, and poor success. Simultaneously, this need of precision therapeutics requires solutions to enable the treatment to be tailored to particular subpopulations of patients. Drug repurposing is a tactical shortcut, making use of the available safety history of pre-existing compounds to expedite the development of therapy. The hypotheses of this review are that convergence of artificial intelligence (AI), big data and systems pharmacology can provide a reinventive, integrative framework that drives this novel paradigm. We describe how AI-based models, which are based on systems-level network analysis, can be used to predict new drug-disease relationships in a systematic way- not by chance but by hypothesis-guided precision repurposing. The discussion includes major pillars of methodology, such as signature-based matching and knowledge graph reasoning to deep learning on biological networks. Using illustrative case studies in oncology, rare diseases and pandemic response, we show how an integrative AI workflow, in the context of candidate prioritization and mechanistic elucidation, is operationalized. The achievement of this potential, however, depends on addressing the major challenges, such as data heterogeneity, limitations of algorithms as black boxes, and translation problems in validation and regulation. Finally, AI-based systems pharmacology will be a paradigm shift of more efficient, guided, and patient-centric therapeutic discovery.