“Deep learning” is an advanced machine learning technology that enables the generation of relevant insights from enormous amounts of data. Computers are able to identify complex patterns in data from many disparate sources and generate algorithms to explain them. This allows a computer to simultaneously create and test original hypotheses with much greater speed and accuracy than a human researcher can. The result is a faster, less expensive discovery process that is proving itself with a growing number of drug developers in terms of increased R&D effectiveness and lower risk. While we are not aware of any drugs identified or developed via AI strategies yet reaching the market, a number of compounds are now in various stages of clinical testing.
Such applications of artificial intelligence (AI) were initially employed by a number of large pharmaceutical firms and start-up companies in the area of drug repurposing — looking for new applications of existing drugs or drugs that initially failed in the indications for which they were originally studied. But increasingly, these computer-learning methods are being used across the drug R&D process, from early-stage discovery research to late-stage clinical trials, and across such functions as virtual screening, lead identification and optimization, biomarker identification, preclinical research, toxicology, clinical trial support, and patient recruitment. GlaxoSmithKline, Sanofi, Merck, and Novartis have all been exploring the potential of AI-driven drug discovery, primarily through collaborations with start-ups specializing in defined areas of application for such technology. Major technology giants, including Microsoft and IBM, are also investing in applications of AI for drug design and pharmacology.
Most recently, Benevolent AI — Europe’s largest AI company and one of the top five AI institutions worldwide — has made a move that makes it the first AI company capable of the full spectrum of internal drug R&D. In February 2018, Benevolent AI acquired the former Proximagen’s UK drug discovery and development center in Cambridge, England. Prior to this acquisition, Benevolent AI could only conduct discovery efforts in-house and relied on Contract Research Organizations and collaborations with research institutions for its drug development activities. Now the company is combining its AI prowess with a large scientific team, led by former GSK Head of Discovery Performance Dr. Ian Churcher, having expertise in assay development and screening, medicinal and synthetic chemistry, drug metabolism and pharmacokinetics, pharmacology, and clinical development. The company currently has 20 research programs underway in areas including rare cancers, inflammation, neurodegenerative diseases, and other central nervous system disorders. Benevolent AI is betting that it can cut the costs associated with drug R&D by 60% and reduce the timeframe for drug design from three years to only one.
Will the application of AI technologies to drug discovery have a major effect on productivity and innovation within the pharmaceutical industry? Will other AI firms decide to invest in their own drug development teams or will they remain as partners to pharma and biotech? And will CROs add AI capabilities to their own offerings, as this technology increasingly enters the pharmaceutical R&D service space?