By: Manasa Kalaimalai
Everyday, as situations around the globe worsen and the world grapples with COVID-19, the use of technological innovations has been more incessant than ever as we struggle to find a solution to the crisis at hand. Although we speak of artificial intelligence being the future, it seems as though the future is already here as artificial intelligence and machine learning have the potential to play a significant role in helping researchers and scientists better understand the current pandemic. But how exactly could they be implemented in the current COVID-19 situation? Several approaches are seen below.
Risk Prediction for Covid-19
Machine learning has proven to be an extremely invaluable asset when it comes to risk prediction. In terms of the current pandemic, it is being used in mainly 3 ways :
1. Infection Risk (individual risk of getting COVID-19)
2. Severity Risk (risk of developing severe complication of COVID-19)
3. Outcome Risk (risk for treatment to be ineffective and risk of death)
In order to do the above, researchers are developing a multi-source deep neural network-based predictive tool to combine demographics policies, regional infections, and individual risk information for accurate risk evaluation.
Diagnosis of COVID-19
Testing patients on a large scale not poses challenges regarding safety, but is also very costly, especially in the earlier stages. Considering that the symptoms of COVID-19 are quite common- cold, cough, sore throat- many patients are bound to be concerned that they might have contracted the disease if they come across any of the symptoms. This causes testing processes to become much slower and costlier. This is where machine learning comes in to simplify, quicken, and make the tests cheaper.
Upon entering a hospital, patients in a Florida hospital were to undergo a face scan that uses machine learning to detect whether or not they have a fever. Doing so helped aid nurses triage between patients effectively.
Furthermore, many doctors nowadays are dealing with 20-hour work shifts with an influx of patients every continual day. A significant amount of their time is spent answering the basic questions of distraught patients, whereas they could be focusing on patients facing more dire circumstances. In this case, chatbots- acting as self-triage systems- could be implemented wherein people self-identify the best course of action to take based on their given symptoms. Large companies, such as Microsoft, have already begun releasing these chatbots for public use.
Deep learning techniques under machine learning can prove to be extremely indispensable considering how critical it is to develop a vaccine as soon as possible. Although current methods involve a tedious and long process of trial and error, machine learning could speed up this process significantly using supervised learning methods to narrow down on several potential drug compounds that have potential to be a vaccine. An investigation regarding the same idea was conducted by Benevolent AI, a UK AI company, where within days they found that Baricitinab ( a drug used for rheumatoid arthritis) proved to be the most prospective candidate for a vaccine. Baricitinab is currently in late-stage critical trials.
Clearly, machine learning has the potential to solve the global pandemic that billions are struggling to push through. It is truly amazing how quickly organizations are implementing machine learning methods to address COVID-19 and aim to bring to fruition the promises that technological innovations in artificial intelligence were to fulfill for future crises. Now only time can tell whether or not machine learning will truly be able to successfully solve in the current calamity we’re facing or whether its promising applications are just overhyped.