As we head into the future, we are trying to rely more on machines to get the job done. But making machines work seemed to be a herculean task until the concept of Machine Learning was born. Machine learning is an art of enabling a machine to make decisions of its own without explicitly programming it. In layman terms, it is the way of teaching computers to learn by itself. The way it works is by understanding the pattern in a given data and making a prediction based on the same. For instance, if you feed the computer with different images of cats and dogs, the computer would aptly differentiate between them based on the physical characteristics of Cat and Dog acquired from the data. The application of Machine Learning is not just limited to this cat & dog problem. Its versatility makes it employable on a wide variety of problems that we face in daily life. For instance, your favorite social media platform, Instagram uses Machine learning to identify those accounts which are similar to one other.
As the entire world is battling the novel coronavirus, Scientists have developed a way to aid the doctors in detecting coronavirus easily and, most importantly, quickly. As of now, the healthcare professionals are testing for coronavirus by extracting cheek swab or other bodily fluid. The downside of this is the amount of time it takes for the patient to receive their results as well as the risk of giving away false-negative results. On average, it takes around 1-2 days to receive the results but as the infection rate has increased and the number of people getting tests has gone up, it takes as much as 4 days to get the results back.
Machine Learning is a viable solution for this. If all turns out well, the patient can receive their test result in about 20-30 minutes. The way it works is relying on CT Scans than fluid tests. When a person is infected with novel coronavirus, their lungs show an abnormality in the scan results. The reason for this is that, when the virus reaches the lungs, the immune system of the body would respond and start fighting against it. Though the immune cells kill the virus, it also affects the healthy cells present in the lungs. This alteration would subsequently result in the accumulation of fluid. This is known as ground-glass opacity and it is the reason for white patches in the test results.
This abnormality in CT Scan results would give the doctors an indication that the patient is infected with the coronavirus. As of now, 3 Million people were tested positive for coronavirus along with 230,000 deaths.
If we are able to get hold of the CT Scan results of these 3 Million patients, then we can feed the data to the computer and teach it how to differentiate between a positive case and a negative case. Once it learns the difference, it will begin screening the results for flagging positive or negative, based on the accumulation of fluid in the lungs. The way they validate their results is by splitting the data into 2 parts for teaching their computer with one while testing using the other (2 Million for teaching and 1 Million for testing). This would give the scientists an idea of the accuracy.
Though training 3 Million worth of data will take ages (like how it takes a lot of time for copying a 1 GB file as compared to 1 MB), Scientists are experimenting with different ways to shorten this time. This is not at all fiction as many scientists have already published research papers on this. In the dire need for more tests, Machine learning can scale-up screening as there is no requirement for a test kit, as a CT Scan alone would suffice. This piece of technology is also providing doctors with a secure mode for screening as the chance for direct contact with the patients is low, as compared to the swab tests. If we are to fight the battle with one of the most dangerous pathogens, then it is always safe to have a machine assisting you! The closeness betwixt the man and the machine can’t get any better than this. In a while, the machine will be aiding the man fighting for his life.