Pharmaceuticals

Artificial intelligence is changing drug discovery

By Sarah Harding, PhD

Artificial intelligence (AI) is set to revolutionize the pharmaceutical industry, from discovery to commercialization, says Sarah

Artificial intelligence (AI) is set to revolutionize the pharmaceutical industry, from discovery to commercialization, says Sarah Harding.
 
Cited by Forbes as being one of the top healthcare predictions for 2019, the long-anticipated era of artificial intelligence (AI) in healthcare is finally taking off this year, with the market for these applications predicted to exceed $1.7 billion.
AI in pharma 2019-04.jpg

For the general public, most interest has been focussed on the way in which AI could address the world’s health problems, and how its use of data from digital health trackers, for example, may help patients to become their own doctors. It has given rise to concerns that AI may replace professionally trained doctors in the future, although it would seem far more likely that AI will simply make doctors better – so perhaps doctors who are using AI will replace those who don’t.
 
Within the pharmaceutical industry, 2019 has seen a burgeoning of interest in the role of AI in drug discovery and development. Pattern recognition, machine learning and AI all play an increasingly important role in rational drug design, screening and identification of candidate molecules and studies on quantitative structure-activity relationships (QSAR).
 
As well as increasing enterprise efficiency with data collation and other ‘digital’ tasks, AI is expected to help companies leverage their assets by finding new targets for existing molecules, or identifying sub-populations that might respond to drugs that initially failed to reach sufficiently high response rates in an overall patient population. It is also anticipated that AI will revolutionize original drug development, as programs will be able to ‘create’ new molecules from scratch and test them against computational models to make predictions of success. The estimated $2.6 billion price tag of developing a new treatment includes money spent on the nine out of ten candidates that fail somewhere between phase I and regulatory approval. If AI can remove the nine failing candidates from that equation, the cost of developing a new drug could be markedly reduced.
 
While all this is still early technology, companies are already starting to invest in this type of approach, and it is expected that AI will evolve rapidly over the next few years, especially in drug discovery and risk analytics applications. By July this year, it was estimated that as many as 141 start-ups are already using AI in drug discovery,1 to aggregate and synthesize information, understand disease mechanisms, establish biomarkers, generate data and models, repurpose existing drugs, design new drugs, generate novel drug candidates, design preclinical and clinical trials, and analyse real-world evidence. That small start-ups are driving this revolution comes as no surprise. It reflects the current industry profile in which – in a bid to minimize risk to big pharma – smaller, creative companies end up funding early-stage innovation.
 
In a recent example, GT Healthcare and associated investors launched a new AI-driven drug development company called GT Apeiron Therapeutics. At the same time, the company entered into a drug discovery collaboration with Exscientia, which uses AI for drug discovery. Working with Exscientia’s state-of-the-art automated AI drug discovery platform and other emerging technologies, Apeiron aims to accelerate the development of novel drugs via new pathways to tackle high impact therapeutic targets, with an initial focus on oncology.
 
Also earlier this year, Merck announced a collaboration agreement with Iktos, for the use of its generative modelling AI technology, which automatically designs virtual novel molecules that have desired activities for treating a given disease. This tackles one of the key challenges in drug design: rapid identification of molecules which simultaneously satisfy multiple drug-like criteria for clinical testing. This followed the announcement, in December 2018, of a year-long licencing agreement between Merck and Cyclica for the use of its AI-augmented proteome screening platform, Ligand Express. Merck also recently announced that it has been granted a US Patent for a novel combination of AI and blockchain technology, aimed at providing a solution for the secure integration of physical products into the digital world.
 
Further downstream, Sanofi and Google have joined forces to develop a virtual healthcare innovation lab, which focusses on three key objectives: to better understand patients and diseases, to increase operational efficiency, and to improve the experience of patients and customers. Sanofi and Google plan to apply AI across diverse datasets to better forecast sales and inform marketing and supply chain efforts. Using AI will take into account real-time information as well as geographic, logistic and manufacturing constraints to help the accuracy of these complex activities.
 
If these and other proponents of AI are right, such technologies will usher in an era of quicker, cheaper and more-effective drug discovery. AI is set to revolutionize the pharmaceutical industry, from discovery to commercialization, leading to the contemplation that – similar to the doctors mentioned above – eventually, it seems feasible that pharma companies that are using AI could simply replace those that don’t.
 
Reference:
  1. Smith S. , July 2019 (https://blog.benchsci.com).