Glaxo hails robot breakthrough

Now artificial intelligence slashes years off time taken to trial drugs

PHARMA giant GlaxoSmithKline is hailing a breakthrough in artificial intelligence technology which could slash years from the time it takes to make new drugs.

GSK is harnessing the power of supercomputers and cutting- edge ‘machine learning’ techniques to discover new treatments.

Currently, it can take up to 12 years and about £1.1bn to develop a new drug, with vast amounts of trial and error involved. But new artificially intelligent robots, which can examine decades of test data to predict how molecules will react, are on the brink of slashing development times.

In addition, the computers will also be used to sift through thousands of previous drug trials which have been deemed flops in the hope of spotting potentially life-saving treatments that may have been missed.

GSK and its British arch-rival Astra Zeneca are already using the technology, with both teaming up with smaller tech start-ups to bring in expertise.

At GSK’s base in Stevenage a team of scientists is already making up to 20 potential drug treatments each month that have been suggested by machines.

It is thought that speeding up development with this innovation is a central part of new chief executive Emma Walmsley’s drive to reclaim GSK’s status as a medical pioneer.

Darren Green, director of computational chemistry at GSK, said machines were on the verge of helping cut the amount of time spent on initial discovery research from five years to just one.

He told the Mail: ‘What AI can help most with is efficiency because, contrary to what you might think, humans are not so great at designing drugs. We have a very high failure rate.’

It takes about ten years for a medicinal chemist to train, gain expertise and build professional judgment, but Green believes machines can reduce the burden on scientists. He said: ‘Humans are not perfect... but the machine can be the chemist with the perfect memory and help draw on all those sources of information.

‘So when we get a problem, we can say, “Have we seen this before? And when we did, what did we do? What worked, and what didn’t work?”

‘We are still a long way from the machine doing it all, but it is at a point where it can definitely improve efficiency, which means getting things to patients faster and making drugs cheaper.’

GSK has been working with Dundee technology company Exscientia, which specialises in using artificial intelligence.

In a deal worth £33m, the firm is looking to find drug treatments on ten projects.