Content area
Full Text
Machine learning is a fundamental concept of artificial intelligence (AI), and is a key component of the ongoing big data revolution that is transforming biomedicine and healthcare (1-3). Unlike many ‘expert system’-based methods in medicine that rely on sets of predefined rules about the domain, machine learning algorithms learn these rules from data, benefiting directly from the detail contained in large, complex and heterogeneous datasets (4). Deep learning is one of the most successful types of machine learning techniques that has transformed many important subfields of AI over the last decade. Examples include data modeling and analytics, computer vision, speech recognition and natural language processing (NLP). Deep learning demonstrated breakthrough performance improvements over pre-existing techniques on a wide range of complex tasks across multiple biomedical research domains spanning from basic clinical to translational (5). The deep learning methods landscape encompasses a variety of biologically inspired models that can be applied directly to raw data, automatically learn useful features and make a prediction without a need to form a hypothesis (5). While biomedical applications of deep learning are still emerging, they have already shown promising advances over the prior state-of-the-art in several tasks (6-8). We anticipate deep learning algorithms to have a substantial impact on pharmacogenomics, pharmaceutical discovery and more generally, on personalized clinical decision support in the near future. Pharmacogenomics focuses on the identification of genetic variants that are correlated with drug effects in populations, cohorts and individual patients. It has traditionally straddled the intersection of genomics and pharmacology, with the greatest impact on clinical practice in oncology (9), psychiatry (10), neurology (11) and cardiology (12). Pharmacogenomics offers promise for applications such as medication optimization for patients based on genotype in diagnostic testing, value as a companion diagnostic and drug discovery and development. However, physicians, caregivers, patients and pharmaceutical and biotechnology companies have all been slow to adopt pharmacogenomics, despite recommendations by the US FDA (13). Recently, however, pharmaceutical companies that are faced with rising costs and resource investments required for drug development have begun to recognize the potential of genomics for drug discovery, and to a lesser extent, for stratification of participants in clinical trials to mitigate adverse events (AEs) and increase efficacy (14,15). In addition, the adoption of pharmacogenomic testing for optimization of...