Can AI help fight covid-19?
A new book, written by leading doctors, explains how cutting-edge research can help fight viruses
To find future treatments for infections, we might have to look back at the past. Before the discovery and advent of antibiotics, doctors had an altogether different strategy to treat bacterial infections. The treatment is called phage therapy, and it uses special viruses called bacteriophages or phages for short. Phages are harmless to humans, but they infect and kill bacteria. In 1923, the George Eliava Institute in Tbilisi, Georgia, pioneered the use of phages to treat human infections. The very same treatment from the 1920s is now being used to treat modern bacterial infections resistant to many drugs. The problem with repeatedly using the same antibiotics is that bacteria eventually evolve to become resistant to them. However, when bacteria evolve to beat a phage, the phage virus also evolves to once again kill the bacteria!
Phages and bacteria are locked in an ever-ongoing arms race with each other. As more antibiotics are failing us, scientists are turning to phages, which have been engineered by nature’s wisdom to stay a step ahead of bacteria. In 2019, phage therapy cured two patients with drug-resistant infections. One patient was about to have his leg amputated due to an MDR (multi-drug resistant) bacterial infection. Doctors used an experimental phage therapy that cured him and saved his leg. Another patient was comatose and near death due to a MDR bacterial infection. When all antibiotics failed, doctors at University of California, San Diego, used an experimental phage therapy to save his life.
Viruses have also been used to treat diseases other than infections. Virotherapy is a biotechnology-based treatment that reprogrammes viruses into therapeutic agents to treat diseases. In 2015, Talimogene laherparepvec, a genetically engineered herpesvirus, was approved by the (US) FDA to treat inoperable melanoma, a type of skin cancer. In 2017, Luxturna, a virus-based gene therapy, was approved to treat retinal dystrophy in adults. In 2019, Zolgensma, another virus-based gene therapy, was approved by the FDA to treat spinal muscular atrophy in children. At a cost of over $2 million per patient, it is the world’s most expensive treatment.
Numerous other virus-based treatments for cancers, genetic disorders, immunodeficiencies and infections are being developed, and the future of virotherapy is bright. Perhaps, future drug discovery need not be limited to serendipity and human intelligence. We are entering a new age in the battle against disease, the age of drug development powered by viruses, artificial intelligence and machine learning.
Recently scientists added a new potential antimicrobial agent to their arsenal, designed by artificial intelligence. In February 2020, scientists at Massachusetts Institute of Technology published exciting new research. They trained an AI deep-learning model to predict and develop antibiotics. AI found a broad-spectrum antibiotic that was lethal for numerous MDR bacteria. Scientists named the new drug Halicin after Hal, from 2001: A Space Odyssey. Besides its broad spectrum, Halicin is also resistant to future bacterial resistance. Halicin targets such a fundamental part of bacteria that they are likely to need at least two or more sequential mutations to overcome the drug, something that is less likely. While the drug is still to be tested in humans, in mice it rapidly cleared up severe infections due to MDR bacteria.
Turning our attention back to coronaviruses, SARS-CoV and SARS-CoV-2 rapidly spread across continents and used similar strategies to infect humans. The tactical response to contain COVID-19 has also been similar to that of SARS, but one major difference exists. In the nearly two decades since SARS, AI could be instrumental in the fight against the virus: A Canadian AI company, Blue Dot, was amongst the first organizations to reveal the COVID-19 outbreak in late December. Another AI company, BenevolentAI, discovered that a drug approved for rheumatoid arthritis, baricitinib, might be effective for COVID-19. Insilico Medicine in Hong Kong used AI algorithms to design six new molecules that could halt viral replication.
While China is the epicentre of the outbreak, it is also the country playing a leading role in developing and using AI to battle the virus. Infervision, a Beijing-based AI company, can diagnose COVID-19 using CT scans in just ten seconds. Manually reading a CT scan can take up to fifteen minutes, and during this time doctors, hospital staff and other patients are at risk of being infected by SARS-CoV-2. AI and machine learning are also being used to model how infectious diseases spiral into major outbreaks and become pandemics. Researchers are using these models to predict how COVID-19 will spread so that they can be better prepared to fight it. While it is premature to gauge to what extent AI will affect the COVID-19 outbreak, AI will probably play a role in containing this outbreak and even more so in future outbreaks.
Excerpted and edited from The Corona Virus: What You Need To Know About The Global Pandemic, published by Penguin Random House India this week.