IIIT Delhi develops AI model to find a drug for COVID-19 from existing drugs



Three organisations joined hands to present an AI algorithm to find drugs that have the highest probability of success against COVID-19 from all available medicine.

Share this Article:

The Delhi-based Indraprastha Institute of Information Technology (IIIT) along with Institute of Post-Graduate Medical Education and Research (IPGME&R), Kolkata and the Inria Saclay-Île-de-France Research Centre, France has worked on an Artificial Intelligence (AI) algorithm that can help quicken the search for COVID-19 treatment. The AI model will work on repositioning the treatment of COVID-19 by checking for a medicine which has the highest probability of effectiveness against the COVID-19 virus. 

Usually, such a research would have taken months, if not years, if researchers would have attempted to do so manually. Parallelly, the search for a new drug will take time too, as any new drug needs to be tested through multiple trials before they are approved by national and international agencies. Similarly, the discovery of a vaccine against COVID-19 seems at least a year or two away too.

In such a scenario, repurposing available drugs for the disease seems to be the only timely solution. In the recent past, drugs such as Hydroxychloroquine (HCQ), Dexamethasone, Remdesivir, Avifavir/Favipiravir etc. have been repositioned for COVID-19. The other drugs which are on trial include Ribavirin, Umifenovir, Sofosbuvir and several antiretrovirals.

The AI algorithm created by the trio prune through a database of drugs called drugback.ca which lists more than 100 antivirals that have been approved. The AI algorithm will follow the drug-target interaction (DTI) prediction approach on this list to discover which drug, originally meant for a different ailment, can be repurposed to treat a certain disease. 

The AI algorithm will first identify similarities between the genomic structures of the existing viruses and the novel coronavirus, and then compute the similarities between the chemical structures of the drugs. For the next step, the algorithm will look at the historical information regarding the effectiveness of the drug on different viruses. Based on this information, drugs that were effective in treating viruses with a similar genomic structure as the novel coronavirus will be selected. 

Such deductions have worked in the past. For example, the drug Imatinib mesylate was originally discovered for leukaemia treatment but was proven effective in treating gastrointestinal stromal tumours too. 

One constant challenge for doctors and epidemiologists is that the novel Coronavirus is constantly mutating at a fast pace. For example, drugs that were effective on the strain of novel coronavirus in December 2019 are very different than the drugs effective on the strain in June 2020. Thus, AI algorithms like the one developed by IIIT-Delhi can help professionals evolve the drugs based on the evolution of the virus. 


Image by nir_design from Pixabay 

Previous News

IIT-Roorkee offers 'AI in banking' course

Next News

Researchers suggest a framework to measure AI’s impact

Suggested Articles