The severity of the ongoing coronavirus pandemic has led to a frantic search for not just a cure, but also viable treatments as well as better ways of analysing the effects and spread of the Coronavirus. Amongst the tools being harnessed for the fight against COVID-19 is Artificial Intelligence: It was an AI epidemic monitoring tool that first raised the alarm over a new viral infection emerging in Wuhan, China. Soon after that, Google’s DeepMind system came up with predictions of protein structures associated with this hitherto novel virus. Meanwhile, Google’s COVID-19 Research Explorer uses NLU to help medical personnel access relevant data without having to sift through thousands of studies. Then there’s the COVID-19 Open Research Dataset, which makes raw data publicly available for AI teams in the hope some team, somewhere, could glean lifesaving information from the incoming flood of data. 

This shouldn't come as a surprise. We rely on AI and ML to make sense of immense amounts of information, and that’s what the Coronavirus epidemic has thrown at us - tonnes of data on the spread, the infection mechanism, the human body’s response, the effect of medication. But there’s one more area where AI’s contribution could make a difference: the search for effective drugs that could open the door to new treatments, and perhaps, even a cure.

AI can help slash development time and assist in repurposing existing drugs

Artificial Intelligence has long been used for drug discovery, with one market study estimating that the AI drug discovery market could touch $1.4 billion by 2024. One of the more prominent examples is that of Pfizer, which had partnered with IBM to use its Watson platform as far back as 2016. The reasons for adopting AI for use in pharma research are many, but one that’s relevant to the race-against-time position we find ourselves in is that AI can help slash development times drastically, by helping ‘filter’ drug candidates without costly (and time-consuming) modelling or testing. 

Today, the headlines belong to startups like Exscientia, which has previously worked with pharma majors such as Sanofi. Exscientia, which describes itself as a ‘full stack AI drug discovery company’, has now turned its attention to the Coronavirus with a partnership with drug research nonprofit Scripps Research - they’ll be using their AI platform to scour Scripps Research’s library of compounds (including existing drugs) for use against the SARS-CoV-2 virus. 

But Exscientia is not the only AI-focused pharma company looking at drug discovery for COVID-19. There’s Benevolent.ai, which is involved in everything from using AI to develop new insights into diseases, to repurposing existing drugs for tackling other diseases, to AI-backed new drug development. Benevolent’s work on COVID has already shown results, with the drug Baricitinib (used for rheumatoid arthritis) now undergoing trials to ascertain its efficacy in treating severe cases of COVID-19. Another pioneer from the AI field is Healx, which in normal times, was focused on rare diseases, but is now using its Healnet drug discovery platform to come up with drug combinations that could help combat COVID. 

Argonne National Laboratory, run by the US government, is yet another scientific institution that has turned its attention towards COVID, and amongst its efforts is a programme that uses AI to screen and filter promising compounds scientists can explore in greater detail 

But perhaps the best example of how AI helping could make a difference during this pandemic is that of data-driven drug research firm AI Vivo, which had, in April, identified the steroid dexamethasone as a potentially useful drug in battling COVID. AI Vivo’s platform first generated a model of the COVID virus, identified targets, and then analysed 90,000 drugs over a period of 15 days to come up with a shortlist of compounds. Today, following clinical trials, Dexamethasone has been approved for treating COVID-19 by the UK NHS.

Vaccine researchers are using AI

But the holy grail for COVID treatments is a vaccine, which might very well be the only way to eradicate a disease that has killed nearly half a million people across the world. Vaccine development is a slow process, one that takes years, if not decades, but researchers are using AI to cut the development time by modelling new compounds as well as identifying the best way to attack the virus: Bengaluru-based startup Sravathi Advance Process Technologies is using AI models to generate new molecules that may one day form the basis for a vaccine, while the Australian e-Health Research Centre is reportedly using AI tools to determine the most likely mechanism of action for any vaccine. 

Harvard University and the Université de Montréal have also launched a similar effort (Epitopes.world) to predict the best ways of attacking the Coronavirus. In their own words, “Testing vaccines is a very lengthy process and patient samples are rare, fragile and precious. By making our results public, our goal is to significantly reduce the pool of targets to be tested, thus accelerating the development of a vaccine.”