Pharma & drug development Science & tech

AI drives discovery

British artificial intelligence company BenevolentAI’s recent license agreement, for a series of novel clinical stage drug candidates, could hail a significant breakthrough in the way drugs are developed. Abi Millar finds out more.

In November 2016, the British artificial intelligence company BenevolentAI announced it had moved one step closer to achieving its goals. It had signed an exclusive license with Janssen Pharmaceutical, a Johnson & Johnson company, giving it sole rights over a series of novel clinical stage drug candidates and patents. This will enable the company to deepen and accelerate its development pipeline.

“This was a milestone for our company, unleashing a wealth of data that we can now evaluate using our AI technology,” says Professor Jackie Hunter, board director of BenevolentAI and CEO of Benevolent Bio. “We can now determine the potential of these small molecule compounds to be useful medicines and are looking to develop them for diseases with high unmet medical need.”

Reinventing methods

Although it is not yet clear which diseases these compounds will be targeting, Benevolent Bio currently focuses on inflammation, neurodegeneration, rare cancers and ‘orphan diseases’ such as Amyotrophic Lateral Sclerosis (ALS). The company has said it aims to remain as flexible as possible, moving into additional areas of biomedical discovery should the right opportunity present itself.

Under the terms of the agreement with Janssen, it will have the sole right to develop, manufacture and commercialise the new drug candidates in all indications, and across all regions. The development process is already well underway, with late stage Phase IIa clinical trials expected to begin in mid-2017.

While this kind of agreement would be cause for celebration at any biotech company, BenevolentAI is no ordinary biotech company. Rather than abiding by the usual mores of medical research, it aims to change the very way knowledge is created.

“In a way we want to reinvent the scientific method for the 21st Century by turning the world’s mass of highly fragmented scientific research data in new insight and usable knowledge that benefits society,” Hunter explains.

Accelerating drug discovery

BenevolentAI was founded in 2013, with the goal of harnessing artificial intelligence to drive rapid scientific discovery. Headquartered in the ‘Knowledge Quarter’ tech cluster in Kings Cross, London, it sits alongside the likes of Google and Facebook, where no ambition feels too remote.

Its two branches, Benevolent Bio and Benevolent Tech, have slightly different aims: the former, to apply the company’s tech in human health and biosciences; the latter, to develop that tech for use in multiple industry sectors. In both cases, the intention is to bring the power of humans and machines together.

“BenevolentAI is applying AI and deep learning techniques to enable the analysis of vast quantities of complex scientific information,” says Hunter. “We’re harnessing AI to accelerate drug discovery, helping our – and other – scientists make sense of and draw new insights from the world’s mass of scientific data.”

The need for such technology is not in doubt. Despite huge leaps in knowledge, the tools of scientific discovery have barely changed in 50 years. With medical information growing exponentially, traditional methods of analysis are poorly equipped for today’s research landscape.

“It is impossible for scientific researchers to process all of the information potentially available to them – in life sciences alone a new scientific paper is published every 30 seconds and there are thousands of updates to PubMed every day,” says Hunter. “Consequently, only a small fraction of globally generated scientific information can form ‘useable’ new knowledge.”

AI promises a solution to this conundrum. Through taking in huge amounts of data, AI can be taught to home in on important-seeming information, develop hypotheses, and refine its abilities over time. Over the last few years, as startups like BenevolentAI have got off the ground, the industry buzz has been palpable.

“AI essentially allows us to analyse the sum of all scientific knowledge, a vast amount of scientific information, and to create new, usable and deep knowledge for researchers. This allows science to access insights more quickly and effectively, in a way it is simply not able to do with existing technologies,” says Hunter.

Clearly, this can speed up the process of finding new targets. But it also dramatically improves the odds that a candidate will prove efficacious. In short, because AI can home in on the mechanism by which a drug acts, there is less trial and error involved. This in turn can significantly cut the cost of drug development, speeding up the passage of drugs to market.

Joining the dots 

BenevolentAI uses a technology called a Judgement Augmented Correlation System (JACS). Through employing cutting-edge deep learning algorithms and natural language processing, this scours masses of scientific articles and databases. It then homes in on relationships between disparate information sources, forming useable insights for researchers.

In the case of ALS, for instance, the AI reviewed billions of sentences from millions of scientific research papers and abstracts.

“JACS began to link direct relationships between the data and regulated the data into ‘known facts’,” says Hunter. “We then curated these known facts and used them to develop new connections and relationships, which enabled us to develop 200 novel hypotheses. Our expert scientific teams were then able to assess the validity of these hypotheses and prioritised five approaches which are now being tested in biological experiments.”

The drug candidates acquired from Janssen were selected through similar means. Through leveraging their AI technology, the team was able to evaluate many small molecule compounds, ultimately picking those that hold the most promise.

“We now have a very strong clinical pipeline with a number of drug candidates expected to enter Phase II and Phase I next year – progress that previously may have taken a biotech at least 10 years to achieve,” says Hunter.

In the expensive, resource-intensive world of drug discovery, it is hardly surprising that AI has been touted as a gamechanger. BenevolentAI, in fact, plans to go further still, expanding into other scientific industries such as veterinary medicine, agrotech, nutraceuticals and materials science. In this light, its recent agreement with Janssen is just a stepping stone.

“Taking this technology to scientific fields like drug discovery allows us to accelerate the development of knowledge enabling human experts to access more data, of higher quality in a faster way,” says Hunter. “This has the potential to revolutionise the scientific process, and can go beyond ALS and even drug discovery, solving all kinds of challenges we currently face in the 21st century.”

This article appears in the February 2017 edition of Pharma Technology Focus

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: