Context
COVID pandemic has affected the entire world and the cases are increasing day by day. More than 22 million cases have been reported across the globe and there have been more than 0.75 million deaths. The difficulty is compounded by the fact the virus is new and scientists have limited knowledge about how it behaves. The virus has also shown mutations. This has made it very difficult to find the vaccine. The cases have gone up so sharply that the health care systems in various countries including developed countries have collapsed. In this scenario a need is felt to have a decision support system to handle the pandemic. Machine Learning tool has a great promise to provide the same.
What is Machine Learning?
Machine learning is a software tool which is inspired by the structure and functioning of our brain. Our brain assimilates the image of a particular object and is able to recognize a similar object by matching it with the past image. Similarly, a machine learning tool trains itself with a known data and later makes predictions and inferences about a new data. This expedites decision making which can help in effective COVID management as described below.
Finding a medicine
Machine learning helps us in understanding how molecules interact - something which is critical for identifying new medicine for a disease. COVID virus has a particular RNA sequence which responds differently to different chemical compound or drug. These interactions can be predicted using machine learning tools which saves time in research. Chemical compounds which show favorable predictions may be shortlisted and new drugs can be invented.
The molecular structure of the COVID-19 virus is not fully understood. These structures can also be predicted using machine learning which will helpa scientist find a new vaccine.
Whether anti-viral drug can work for COVID has been a matter of discussion in scientific community. Machine learning is being used to predict the bonding between the anti-viral drug and the protein of COVID-19 virus. This has helped in prescribing a treatment.
Forecasting Infection and spread
Everybody is not equally affected by COVID-19 virus. A study shows that around 82 percent of the infected people show mild symptom while others are severe or critical. The symptomsdepend upon age and other risk factors. A machine learning tool has been developed to predict the response of a patient to the virus. This helps in deciding who should be isolated first.
One of the challenges in COVID is its rapid spread in a geographical area. Significant work has been done for disease forecast of seasonal flu etc. Scientists are trying to use the machine learning tools to predict the pattern of the COVID outbreak. This will help in keeping the hospitals and health staff in place to combat the disease.
Diagnosis and treatment
The virus causes similar symptoms as in normal diseases. For example, pneumonia can be due to a normal health issue or due to COVID. In an x-ray scan both however show different footprints which can be analyzed by a machine learning tool. If the tool is first trained using a large number of past x-ray scan data of COVID patients it can predict whether a new x-ray scan data relates to a COVID patient. This helps in the time taken by radiologists to identify the x-ray scans of patients and treatment can be started early.
Information from social media
COVID has affected our lives immensely. Today social media is full of feedback on the disease. It contains valuable information about its symptoms, spread etc. This can be analysed and used in the management of the disease.
Challenges
While machine learning has great promise in terms of predictions it can make, the challenge is to find the past data which is used to train the algorithm. There is limited COVID related data which is freely available for analysis. It is important that countries across the globe share disease related data openly so that the tool can be developed effectively.
Comentarios