The capabilities of machine learning have been a highlight amid the fight against COVID-19, guiding us not only in our mission to understand the virus, but also in the development of a vaccine.
The world has witnessed arduous efforts from all trades to contain COVID-19’s spread. The pandemic pushed two entities of the globe to work in conjunction, which otherwise would have taken years, even decades, to make progress together: medical research and machine learning (ML).
Why and how of ML
The past year showcased the sheer power of machine learning. While medical researchers have been racing to find a solution for the novel virus, machine learning has been a cornerstone in making meticulous and timely moves in the direction of vaccine production.
ML in learning about the virus
In the early weeks of the pandemic, a team of computer scientists at the Stanford Institute for Human-Centered Artificial Intelligence, led by Russ Altman and Binbin Chenat, used machine learning to study the proteins of the virus. The virus involves a complex structure that needed analytical attention. AI-assisted systems sorted through thousands of complex structures of the virus and investigated crucial parts of the specimen that could be weakened. The investigation resulted in identifying those key pieces of the virus that can be used as antigens.
Immunologists working with ML identified nearly one million protein fragments on a cell’s surface which the human eye could never detect. Several biological arrangements were also discovered in the search, which may have not been possible using manual research methods. In these instances, ML can easily make sense of the data and expose all the hidden bio-chemical patterns.
ML for vaccine design
When vaccine development commenced around the world, the procedures targeted T-cells and B-cells of the human immune system. Typically, for any foreign virus, B-cells produce anti-bodies which disable the pathogen and prevent it from entering healthy cells. T-cells are responsible for destroying cells that have already been infected by the pathogen. The outer membrane of a virus consists of spike proteins that bind with host human cells and start injecting genetic material. A normal human body immune response takes days to weeks to fight any new virus, during which the body becomes weak, and falls sick.
For vaccines, these spike proteins are the main target. In the process of designing a vaccine for COVID-19, the role of AI was to study the complex structure of the virus, learning about its spike proteins in detail and sorting through massive data to determine those parts of the virus that are easy targets for the vaccine. Most importantly, viruses are constantly mutating. AI has been used extensively to identify non-mutating parts of the virus for the vaccine to stay effective for a longer period.
ML for vaccine testing
Clinical vaccine trials for patients that incorporated AI tools have significantly improved trial and testing mechanisms in more ways than one. From identifying the ideal sample size to what data should be experimentally generated, to target validation and trial execution, analytical ML has played a primary role during vaccine testing protocols. ML algorithms have helped in consuming vast data sets, analyzing them, finding patterns in the human response behaviors, and making predictions about future state of the disease.
Additionally, ML has improved the safety and efficacy of trial procedures, accelerating the development of new vaccines. ML has even predicted future outbreaks, providing key insight to which future trials and sites would be most impacted. Such endeavors directly reduce the physical barriers, time and cost of trials.
ML used for vaccine distribution
With such wide populations in each country and even wider segments of communities, determining the distribution of vaccines to those who are at greater health risk becomes challenging. To meet that challenge ML leveraged all the data available to detect important hidden patterns that may otherwise go unnoticed, such as factors about individual health, social and environment conditions, and implications of age.
ML is also helping in strengthening the supply chain for vaccine distribution. “When a vaccine gets distributed at scale and speed, a technology solution needs to track the batch and lot numbers to know exactly where each dose is and who received it,” said Eric Sandor, AI lead at Genpact. The record–related tasks can be completed in minutes, compared to human labor which would take hours.
For a whole year now, the world has been facing the pandemic with grit and patience. With growing solutions to world challenges, Machine Learning has been one of the key players leading the way out.
At Trinus, we are excited to see and experience the new ways technology is helping us beat the global pandemic. COVID-19 has proved ML to be an incredible asset in nearly every industry. Follow us for all the latest news in digital technology, and to learn more about Machine Learning click here.