The global pharma industry is undergoing a dramatic transition from a quest for blockbusters to the design of a precision medicine based drug design. Artificial intelligence is one of the most prominent elements that has been adopted as part of the transition from a fully integrated pharmaceutical company model of drug design to extensive interaction with smaller innovative R&D companies as well as academic institutions.
Artificial Intelligence (AI) is the activity devoted to making machines intelligent, and intelligence is that quality that enables an entity to function appropriately and with foresight in its environment (definition proposed by Nils J. Nillson, Stanford U.). Even though there are numerous definitions for AI, this one fits nicely into the goal of using machine learning for improving the rate of success in the design of novel and cost-effective therapeutics.
One of the primary reasons that AI has such a great potential in drug development is that there is a huge amount of health data available right now in the public health system. Clinical trials’ data, electronic medical records (EMR), genetic profiles and much more is the wealth representing the notion of BIG DATA in healthcare. The main challenge regarding the processing of big data is the need to process it in a meaningful and cost-effective fashion. That is why training a machine to fulfill the task becomes so attractive. Selecting and adjusting the right algorithms is the first essential step but once it is in place, training machines to find optimal patterns between the structure of “druggable molecules” and their optimal activity is within reach.
Canada has established a leadership position in training of machines to learn how to perform complex tasks, in a relatively short period of time. Based on recent commitments to the space, it is expected that we will witness in the foreseeable future designs of novel and much more specific therapeutics with higher potency and lesser side effects. The prospects are quite encouraging in light of the shift global pharma industry is adopting towards precision medicine. That shift will rely on sifting through patients’ medical records. Canadian AI machines are learning fast and are expected to become a key player in advancing academic concepts into standard and streamlined processes and organizations. In Ontario, the University of Toronto has emerged as a world-leading hub for research and entrepreneurship in this area. A potent combination of long-standing academic research in conjunction with the adoption of machine learning methodologies have already proven to be game-changing opportunities. Interactive approaches to computer science and medical research, combined with emerging best in class entrepreneurship programming and training is already yielding some fascinating fruits in the area of AI for drug discovery.
Companies like Structura Bio are taking the complex computational challenge of reducing noisy images from cryo-electron microscopes into readable highly accurate 3D structures of proteins and are doing what used to take a server room filled with computers a week, in a matter of seconds. Similarly, Phenomic AI (a recently incorporated UTEST company) uses a technique called deep learning to analyze data from high-throughput phenomic screens to analyze cell and tissue phenotypes in microscopy data with incredible accuracy. It holds out the potential for eliminating human intervention in the assessment of all that data. In some cases, companies like Deep Genomics and Atomwise are going all the way by leveraging their respective AI technologies to become drug discovery engines themselves. Our awareness of the impact of the AI revolution in drug discovery is already enormous and we’re only at the beginning of its adoption cycle. Future advances in Canada will be buoyed further by strong academic and institutional foundations that have been put in place to assist Canada in sustaining this advantage. The Vector Institute, as an example, was established in 2017 in partnership with Canada’s largest companies and the Federal and Provincial Government’s to attract and retain world-leading research talent and to promote cutting-edge research in the field.
Recently, partnerships have been established between the MaRS Innovation research healthcare ecosystem (UHN, Sickkids, Sunnybrook) with global players in the space of machine learning based drug design and developments. Partnerships with Schrödinger and Evotec have been established to capture the enormous potential of “fishing in the pond” of EMR’s rich source of unraveling the tissue/cellular architecture as a baseline for the discovery of novel disease targets, which thereby establishes a mechanism for better drugs.
The field of AI in the service of medical research is still in its infancy, but the initial avalanche of results is already starting to give us an idea of the great potential that machine learning can offer to those embarking on advancing drug development. Reducing screening times, aiding new drug candidates and finding the most effective drugs for specific diseases at a speed that humans cannot achieve is compelling, and we believe that AI will increasingly become part of the medical landscape. Once hurdles such as data standardized collection and storage as well as data privacy concerned are addressed, it is expected that we will witness an exponential inclination in the implementation of machine learning as a powerful tool in the design of more potent drugs with lesser side-effects. The FDA and Health Canada are encouraging pharmaceutical companies to join the choir.
To conclude, rephrasing from Eric Topol of the Scripps Research Institute (CNBC, May 2017), “The potential of artificial intelligence has probably the biggest impact of any type of technology on healthcare.”
Dr. Hofstein is the President and CEO of MaRS Innovation. He joined in 2009 after past positions that include CEO of Hadasit Bioholding, Israel.