Building trust between AI and pharma industry is vital for success

2 Feb 201866 Shares

Share on FacebookTweet about this on TwitterShare on LinkedInShare on Google+Pin on PinterestShare on RedditEmail this to someone

Image: Pavel Chagochkin/Shutterstock

Share on FacebookTweet about this on TwitterShare on LinkedInShare on Google+Pin on PinterestShare on RedditEmail this to someone

The pharmaceutical industry is on the cusp of an artificially intelligent wave that promises enormous opportunities, but also immense challenges.

The onset of artificial intelligence (AI) in the modern world has led to the prophesying of a seismic shift whereby many of the jobs we are familiar with today will disappear in favour of algorithms and robots.

In the area of pharma, AI promises not only to overhaul how the industry produces and designs medication, but could also lead to the discovery of whole new medication in the space of a few hours – discoveries that would take a human years.

To take just one recent example, scientists at the University of Cambridge announced that, with the help of an AI-powered robot called Eve, they were able to identify a possible antimalarial drug found in a common toothpaste ingredient.

While this is just an example of what relatively small-scale research can do, think about the potential for AI in the vast, mega-pharma corporations with much greater resources at their disposal.

One of those keen to understand how AI can make a real impact in pharma is Dr Muhammed Ali, MSD International’s executive director for healthcare solutions strategy across Europe and Canada in its commercial and operations team.

As one of the upcoming speakers at this month’s BioPharma Ambition in Dublin, Ali has seen over the past few years how the overall rapid advancement of AI systems has contributed to the pharma sector.

Robot scientist

Robot scientist ‘Eve’, based at the University of Manchester. Image: University of Manchester

Improvements from the ground up

Everything from drug discovery, manufacturing, clinical research trial processes and regulatory submission up to product launch, Ali said, has all been subject to automation.

Even at its most basic level, it has shifted administration from large stacks of paper to non-paper documentation.

“Decades of improvement [has led to] analytics that use information to build associations [between concepts] quicker than people poring over graphs in committee rooms making inferences.

“The magic is moving from information to new insights, to actions on where to place enhanced educated bets on success.”

Solving an AI conundrum

But not all technological revolutions are straightforward. Just in the same way there are concerns for where AI will go in the modern workforce, scientists and manufacturers are also facing a dilemma of sorts, where tradition meets new science.

“There’s an interesting cultural conundrum for [the pharma] industry to overcome, and that’s the culture of trust with the science that sits behind AI,” Ali said, adding that the industry “has been built on a robust backbone of ‘traditional’ science that the regulators expect”.

He said that unlike the “rockstar” data scientists of tech giants such as Microsoft, IBM and Alibaba, who are building these systems, the pharma industry has the necessary barrier of regulation and a need to determine that one AI solution is not only feasible for use, but is also safe.

“Other industries are direct-to-consumer so can be driven by demand, but the pharma industry is regulator-led so it requires a naturally more cautious, slower adoption.”

Despite this trepidation, there have already been significant outcomes achieved using AI, especially in the area of personalised medication. Examples include IBM Watson Oncology – the company’s AI that is crunching hard data for individual cancer treatment – being used to great effect.

Or how about MIT’s Clinical Machine Learning Group, which is using machine learning to understand diseases such as type-2 diabetes, and design effective treatments in the process.

Throw on top of that the fact that companies plan to combine AI with other technologies such as the internet of things and blockchain to settle security, privacy and reliability concerns that could arise from smart systems monitoring large numbers of patients.

AI pharma

Image: Kzenon/Shutterstock

Humans in a pharma world

So, should scientists be worried that they too are facing unemployment in the face of increased automation?

Not so, according to Ali, who said that AI systems developed by Google et al are only as good as the scientists and computer scientists who monitor the data and build the systems, respectively.

“AI in its current state is only as good as the data it receives … it isn’t a like-for-like person replacement.”

Rather, he added, scientists will effectively ‘evolve’ to become more integrated with technology than ever before.

“If you think of science-fiction cyborgs who are part-human, part-machine, that’s where we are moving in terms of thinking, rather than anything physical.

“We are about to see augmented learning faster for researchers and scientists who are to be more ‘educated’ at getting to discover, trial and design personalised medicine. The future of AI will force everyone to think AI digital-first and make decisions that are digitally data-driven.”

Dr Muhammed Ali will be speaking as part of the ‘Unlocking the collaborative potential of data and clinical research’ panel at BioPharma Ambition taking place between 21 and 22 February 2018.

Colm Gorey is a journalist with Siliconrepublic.com

editorial@siliconrepublic.com