Machine Learning Helps Combat COVID 19


Clinicians, scientists, and healthcare experts are searching for new technologies to help combat the spread of COVID-19 pandemic. There is significant evidence of Machine Learning (ML) and Artificial Intelligence (AI) applications, having helped in previous epidemics, and researchers have turned their attention to these very same applications to understand how they can help contain the outbreak.

How does Machine Learning Help

ML offers a huge support in detecting the disease with the help of textual and image data. It can also predict the nature of the virus globally. 

Note: ML requires an enormous amount of data for predicting or classifying diseases. Supervised ML algorithms require data that is interpreted for organizing the image or text into separate categories.

Following are a few ways how ML is helping or can help combat the pandemic:

  • Enabling organizations to scale and adjust: ML technology is playing an important role within organizations by providing the tools to facilitate remote communication, enable telemedicine, and secure food. 
  • Understanding how COVID-19 Spreads: ML is helping practitioners and researchers analyse huge volumes of data to predict the spread of COVID-19, to act as an early warning mechanism for future pandemics and to detect susceptible populations.
Note: At the Chan Zuckerberg Biohub in California, researchers have created a model to evaluate the number of COVID-19 infections that go undetected and the consequences to the public health.
  • Identify who is most at risk from COVID-19: ML can be used to predict all three types of risks, which are infection risk, severity risk, and the outcome risk. ML can also be used to predict the risk of detection, who is at risk of developing a severe case, and treatment outcomes. 
Note: AWS has a launched CORD-19 Search – a new search Website powered by ML. This site can help researchers search for research papers and documents quickly and easily.
  • Screening patients and diagnosing COVID-19: Once a new pandemic hits, diagnosing individuals is difficult. Testing on a large scale is tough and tests are expensive. Promising research areas with respect to ML to help diagnose COVID-19, include:
    • Using face scans to recognize symptoms, such as if the patient has a fever.
    • Using wearable technology, such as smart watches to identify obvious patterns in a patient’s resting heart rate.
    • Using ML-powered chatbots to diagnose patients based on self-reported symptoms.
  • Speeding up drug development: With a new pandemic creating havoc, it is critical to come up with a reliable diagnostic method, a drug for treatment, and a vaccine as soon as possible. Current methods comprise a lot of trial and error, which wastes time. ML can speed up the process significantly without compensating on the quality of the drug/vaccine. 
Note: Researchers working on H7N9 found that ML-assisted virtual screening and scoring led to significant improvements in the accuracy of the scores. Thus, they used the random forest algorithm (a classification algorithm made up of various decision trees) that provided the best results with H7N9.
  • Identifying effective existing drugs: There are thousands of drug candidates available, and we do not have enough time to identify and test the right ones. ML can help us identify appropriate drug candidates faster by:
  • Automatically creating knowledge graphs. 
  • Foreseeing interactions between drugs and viral proteins.
  • Forecasting the spread of infectious disease using social networks: Instead of relying on the healthcare system to evaluate the spread of the disease, social media can be used by public to post about their health. These posts can be processed by ML models at scale.
  • Understanding viruses through proteins: ML can also help improve our understanding of viruses by analysing their proteins.
  • Understanding how to attack the virus: Detecting and categorizing epitopes (clusters of amino acids on the outside of a virus) is critical in deciding which part of a molecule to target, once a vaccine is developed. Hidden Markov Model, Support Vector Machines (SVM), and artificial neural networks are efficient in identifying epitopes.
  • Identifying hosts in the natural world and predicting new pandemics: A zoonotic pandemic can survive unobserved for a long time, waiting for the next opportunity to infect us. With ML, researchers can identify potentially zoonotic strains of influenza and its’ hosts accurately, thus helping doctors to predict future pandemics and take appropriate actions.


ML is an essential tool to be used in our fight against the pandemic. If we collate data, sync, and combine our knowledge and skills, we can save many lives.