De-risking Drug Discovery with AI

face
main_img

On the supply side, AI appears to herald the prospect of de-risking the drug discovery process and doing away with many of the traditional time-consuming approaches at a juncture when drug development had been becoming financially unsustainable and clinical research organisations (CROs) had been systematically failing to improve operational efficiency. “There can be no doubt in my mind that artificial intelligence has finally transcended the hype cycle and is now firmly in the mainstream of efforts to advance medical research and drug discovery,” opines Krishna Cheriath, chief data officer at Bristol-Myers Squibb. “One only need look at how innovative pharmaceutical developers are increasingly attuned to the reality that they can secure a competitive advantage by expanding and accelerating the value proposition of their data, analytic and scientific capabilities through AI augmentation,” he notes.

 

Ripe for Disruption

There is certainly an urgent need for the lacklustre attrition rate of pharma R&D to be reversed. Clinical trial outlays over the past two decades have soared – costing an average of $2.6 billion and 12 years to bring a new therapy to market – but the industry has not delivered commensurate high levels of innovation in return,” laments Professor Jackie Hunter, chief executive of clinical and strategic partnerships at BenevolentAI. Indeed, only a mere 13.8% of drugs nowadays actually successfully navigate clinical trials according to the latest analysis from MIT.

“Meanwhile, the past two decades have witnessed an explosion of biomedical data, making it impossible for even the most learned researchers to process and garner real insight from this avalanche of information rendering it all the more imperative to find new tools to harness and utilise that data to optimise the discovery process,” observes Hunter. She believes that AI technology, in its different forms from algorithms and machine learning to deep neural networks, is perfectly suited to this task of mining electronic health records and patient registries to deliver speedy, actionable insights. Others tend to agree, with Novartis CEO, Vasant Narasimhan, boldly declaring that AI-based tech has the potential to excise as much as 20% of existing clinical trial costs.

 

Solving the Pharma R&D Productivity Conundrum

How exactly might such a feat be accomplished? “One of the most obvious use cases for AI in randomized controlled trials is in the identification of patients. In the past, screening patients to include in trials has been problematic, given the mismatch between inclusion and exclusion criteria in electronic health records, but AI can swiftly cut through all of this,” affirms Paul Bleicher, CEO at OptumLabs. Additionally, it can be deployed to assist clinical trial site performance. “Drugs used in clinical trials are typically manufactured in a smaller facility than at launch and AI models can help predict the supply of drugs needed for a trial to facilitate production planning and avoid slowing enrolment because a site runs out of study drug,” asserts Bleicher.

Even more exciting is the role of AI in using biomedical and clinical data to draw unintuitive insights about drug candidates, or even attempting to model the whole biological systems to identify novel pathways, targets and biomarkers. Drug discovery often takes a long time to test compounds against samples of diseased cells. To speed up this screening process, Novartis is already using images from machine learning algorithms to predict which untested compounds might be worth exploring in more details.

There have already been a string of noteworthy successes in this sphere. BenevolentAI and the Canadian outfit, Deep Genomics, have both demonstrated that AI can predict novel therapeutic targets which have then been validated, while companies like Ex-Scientia and In Silico Medicine have shown that AI can significantly shorten the time to find candidate molecules.

Equally the technology opens up new doors to the practice of “drug repurposing” whereby existing molecules are combined in ways that furnish them with therapeutic powers that each lacks in isolation. According to Andrii Buvailois, director of e-commerce at fine chemical supplier, Enamine, “drug repurposing constitutes a particular gold mine for AI-based technologies to drive value since a lot of data is already known about the drug in question.”

 

A Sign of Things to Come

All in all, it is becoming apparent that, when deployed correctly, AI can indeed enable companies to reduce costs and improve success rates in drug discovery and development and that the big winners of tomorrow will likely be those that manage to integrate such techniques. “Powerful forces continue to add pressure and are pushing traditional models of research and development. Companies in the sector must transform to be part of the conversation and relevant or risk being left behind,” concludes David H. Crean, managing director of the investment firm, Objective Capital Partners.

Add Your Comment


You must be logged in to post a comment.

Related Content

Latest Report