It feels as if a superhuman effort is needed to help ease the global pandemic killing so many.
Artificial intelligence may have been hyped – but when it comes to medicine, it already has a proven track record.
So can machine learning rise to this challenge of finding a cure for this terrible disease?
There is no shortage of companies trying to solve the dilemma.
Oxford-based Exscientia, the first to put an AI-discovered drug into human trial, is trawling through 15,000 drugs held by the Scripps research institute, in California.
And Healx, a Cambridge company set up by Viagra co-inventor Dr David Brown, has repurposed its AI system developed to find drugs for rare diseases.
The system is divided into three parts that:
- trawl through all the current literature relating to the disease
- study the DNA and structure of the virus
- consider the suitability of various drugs
Drug discovery has traditionally been slow.
“I have been doing this for 45 years and I have got three drugs to market,” Dr Brown told BBC News.
But AI is proving much faster.
“It has taken several weeks to gather all the data we need and we have even got new information in the last few days, so we are now at a critical mass,” Dr Brown said.
“The algorithms ran over Easter and we will have output for the three methods in the next seven days.”
Healx hopes to turn that information into a list of drug candidates by May and is already in talks with labs to take those predictions into clinical trials.
For those working in the field of AI drug discovery, there are two options when it comes to coronavirus:
- find an entirely new drug but wait a couple of years for it to be approved as safe for use
- repurpose existing drugs
But, Dr Brown said, it was extremely unlikely one single drug would be the answer.
And for Healx, that means detailed analysis of the eight million possible pairs and 10.5 billion triple-drug combinations stemming from the 4,000 approved drugs on the market.
Prof Ara Darzi, director of the Institute of Global Health Innovation, at Imperial College, told BBC News: “AI remains one of our strongest paths to achieve a perceptible solution but there is a fundamental need for high quality, large and clean data sets.
“To date, much of this information has been siloed in individual companies such as big pharma or lost in the intellectual property and old lab space within universities.
“Now more than ever there, is a need to unify these disparate drug discovery data sources to allow AI researchers to apply their novel machine-learning techniques to generate new treatments for Covid-19 as soon as possible.”
In the US, a partnership between Northeastern University’s Barabasi Labs, Harvard Medical School, the Network Science Institute and biotech start-up Scipher Medicine is also on the search for drugs that can quickly be repurposed as Covid-19 treatments.
Normally, just getting them all to work together would take “a year of paperwork”, said Scipher’s chief executive Alif Saleh.
But a series of Zoom calls with a “group of people with a unprecedented determination to get things done, not to mention a lot of time of their hands”, speeded things up.
“The last three weeks would normally take half a year. Everyone dropped everything,” he said.
Already, their research has yielded surprising results, including:
- the suggestion the virus may invade brain tissues, which may explain why some people lose their sense of taste or smell)
- the prediction it may also attack the reproductive system of both men and women
Scipher Medicine combines AI with something it calls network medicine – a method that views a disease via the complex interactions among molecular components.
“A disease phenotype is rarely due to malfunction of one gene or protein on its own – nature is not that simple – but the result of a cascading effect in a network of interactions between several proteins,” Mr Saleh said.
Using network medicine, AI and a fusion of the two has led the consortium to identify 81 potential drugs that could help.
“AI can do a little better, not only looking at higher order correlations but little bits of independent information that traditional network medicine might miss,” said Prof Albert-Laszlo Barabasi.
But AI alone would not have worked, they needed all three approaches.
“Different tools look at different perspectives but together are very powerful” he added.
Some AI companies are already claiming to have isolated drugs that could help.
BenevolentAI has identified Baricitinib, a drug already approved for the treatment of rheumatoid arthritis, as a potential treatment to prevent the virus infecting lung cells.
And it has now entered a controlled trial with the US National Institute of Allergy and Infectious Diseases.
Meanwhile, scientists from South Korea and the US using deep learning to investigate the potential for commercially available antiviral drugs have suggested atazanavir, used to treat Aids, could be a good candidate.
Other companies are using AI for other purposes, such as analysing scans to ease the burden on radiologists and help predict which patients are most likely to need a ventilator.
Chinese technology giant Alibaba, for example, announced an algorithm it says can diagnose cases within 20 seconds, with 96% accuracy.
But some experts warn AI systems are likely to have been trained on data about advanced infections, making them less effective at detecting early signs of the virus.
There needed to be a global effort from policymakers to persuade the big pharmaceutical companies to join forces with smaller drug-data stores, academics and research charities to pool data resources, Prof Darzi said.
“The time has never been more important for drug-discovery data to open up its secrets for AI to help in the battle against Covid-19,” he said.