An outbreak of the latest coronavirus strain to cause acute respiratory disease, COVID-19, has shown artificial intelligence (AI) to be a useful tool in multiple areas of the response effort.
In 2003 the world saw a life-threatening epidemic of severe acute respiratory syndrome (SARS-CoV) and in 2012 middle east respiratory syndrome (MERS-CoV), which continues to be identified in individuals to this day. In contrast to the response to these diseases, AI is being used in various innovative approaches to tackle the outbreak. Canadian analytic company BlueDot, founded by Dr. Kamran Khan who saw the SARS-CoV outbreak in Toronto, has used flight pattern data to successfully predict the migration of the outbreak, able to determine the affected cities and countries long before cases are confirmed.
Mortalities associated with COVID-19 now outnumber those of SARS-CoV, as does the number of worldwide cases; there are confirmed cases in 28 countries as well as on a cruise ship off the coast of Japan. The spread and mortality risk of the virus has generated substantial global interest in development of an effective treatment with the pharmaceutical industry currently funnelling money into antiviral drug development programmes.
However, drug development is a notoriously lengthy process. Before clinical trials even begin the average time taken to develop a drug is 4.5 years, and when it’s a matter of life and death this can be costly in more ways than one. This represents a hugely limiting factor in our ability to control infectious disease.
Four groups have now used AI to identify drug candidates for COVID-19 based on readily available data from previously developed drugs to speed up the development of an effective treatment. Drug repurposing is nothing new to the pharmaceutical industry. When a drug is not successful in treating the disease for which it was developed, it is usually shelved and its data is retained. Over time, pharmaceutical companies accrue a reservoir of drugs and pharmaceutical molecules with associated data resulting from their development and efficacy testing. AI and machine learning are being applied to sift through this vast amount of information to predict which drugs could have reactivity against the new virus based on its genomic data, which was released by China in January.
Atazanavir and Baricitinib, treatments for HIV and rheumatoid arthritis respectively, are among the drugs identified as potential candidates for further testing. The information is offered as a basis for drug development, and whilst the ideal outcome of these studies would be a pre-developed and pre-approved drug ready to deploy, it seems an unlikely result. Nonetheless, drug development for a life-threatening infectious disease is a case of ‘the sooner the better’, so any time saved using AI improves the effect of the eventual treatment.
AI-accelerated drug development has recently enabled a drug for OCD to reach clinical trials in just one year, not only demonstrating its potential effect on the timeline of deploying a treatment for COVID-19 and future infectious disease outbreaks, but also its wider applicability in speeding up the development of any drug.
This post was written by Claudia Carter and edited by Miles Martin