How to fight COVID-19 with machine learning

Covid-19 is one of the ugliest pandemic situations the world has ever seen. As per the latest WHO recordings, more than 21 million people from more than 200 countries have been affected. The death toll stands at more than 750,000 and all the cases are yet not reported. The situation is getting worse by the day. As after-effects of the health situation, is the unprecedented economic meltdown grappling the world over. The past few years have witnessed the growth of technology and its application as an ominous solution for all issues faced by mankind. The question that arises now is that, ‘will technology rise up to this challenge as well?’ The answer is ‘yes’. Machine Learning has proved itself as one of the sharpest world tools in understanding, analyzing, forecasting and in trying to find a solution to this novel virus and issue.

Unwrapping Machine Learning

There is a lot of information and data collected by the various smart machines all across. Machine Learning is basically a subset of Artificial Intelligence, that includes algorithms to sift through this data. The algorithms herein are mostly not man written. Rather, they are machine made as per inputs and outputs of a ‘training data’ batch. The logic amongst them is found by the machine itself and these algorithms evolve as data increases to make more informed decisions.

Using Machine Learning in the Fight of COVID-19

COVID-19 is a new virus on the block. It is caused by SARS-CoV-2 (Severe Acute Respiratory Syndrome- Coronavirus 2) and was first reported in Wuhan, China. The world all over including the Governments, scientists, medicine experts, top mobile app developers and technology enthusiasts are trying to understand the reason behind the spread of this disease and look for logic and ways to cure this disease at their ends. Let us now examine how Machine Learning is being utilized to create our analysis about the virus itself:

Analyzing the risk area Clusters: With the help of apps like the Arogya Setu app in India, and many others worldwide; a lot of data, complete with information about travels, health, is being collected and stored by the Government Servers. In the absence of a proper cure or vaccine, ‘social distancing’ has been predicted as the only effective way ro prevent COVID-19. Machine Learning is being utilized to build robust mechanisms based on mathematical calculations for tracking its spread and automation of tracking tools itself for evolving and dynamic decision making by the machines itself. The data requires bluetooth connectivity at all times and includes most of the viable information about who is suffering from the virus and who is not. Machine Learning is then utilized by processors to sieve through this data to analyze the areas which currently have a lot of victims (risk zones) and which are the safe zones. Since, the major issue of this virus is its prowess to spread indefinitely, machine-learning curated information is helping individuals by informing them as well as the Government about the areas that are required to be cordoned off.

In fact, ML-based improved models are being applied to predict the potential threat of COVID-19 in countries worldwide.

Analyzing higher risk patients: Human body incuscates millions of cells, tissues, proteins and various other intricate structural layers that may respond differently to different viral or bacterial or any other foreign body attack. It is this difference that demarcates the Covid patients that would be asymptotic, those with mild issues and those with acute respiratory and other severe issues with Covid. Machine Learning Algorithms are helping doctors and medical organizations by sieving patient records and mapping data to analyze the ones that are at a greater risk of accentuated SARS symptoms.

Government Reports have analyzed and proven that ‘Machine learning-based risk scores outperformed a widely-used rule-based triage algorithm and human prioritization decisions in predicting hospital outcomes. ’

Machine Learning algorithms are helping to analyze these records logically in terms of age, pre-existing medical conditions, social habits, human interactions, etc. They are then mapped with various treatment options like isolations for the asymptotic patients, medicines for mildly affected ones and the ones requiring higher aid and plasma therapy, amongst others. It has been due to these analyses that Governments have come to a conclusion that heart patients, diabetes patients and other individuals with lower immunity are at a greater coronavirus risk.

Data for developing the Cure and Vaccine: There is unlimited data about the millions of COVID affected patients all across. There is also the research and its various findings that may or may not lead to an effective cure or vaccine for COVID-19. As of now, there are six different types of Coronavirus strains identified by the scientists worldwide. One of the strains is further known to have 2 subdivisions. Thus, there are 7 coronavirus identified strains in all. These strains are further evolving with time.

There is also an area-wise difference in the strains that are spreading. For instance, an article in National Herald India suggests that, ‘The strain of new coronavirus in India is not as virulent as the ones which are devastating Italy, Spain and the US.’ This can cause a major issue in the development of a uniform vaccine for the virus. As vaccines effective on one area strain may not prove to be as effective in other world zones. Herein, machine learning is helping to develop algorithms that shall self evolve to map the evolution process of the virus and then help in generating the possible vaccine outcomes.

Once the vaccine options are developed, another important issue is to analyze the potent candidates for each vaccine’s trials. This can again not be done on an adhoc basis as different patients may carry different background histories, body proteins and immunity. Scientists the world-over have been utilizing machine-learning based tools like Vaxign, which is a reverse- vaccinology tool to zero in on the perfect trial candidates.

Patientcare apart from COVID: Coronavirus is an extremely fast spreading virus form. With an unprecedented number of patients being diagnosed corona-positive, the hospitals have been swarming with patients. Governments have had to convert various city hospitals into COVID centers. The doctors and other medical staff are all overworked and overstressed due to the amount of work required and the risks they are subjected to on a daily basis. But, this in no way means that the other patients (with other diseases) are cured. They too require medical care and guidance about their conditions. Machine-Learning based chatbots to this end are being utilized by medical organizations to calm harried patients suffering from various other diseases like a simple cold, cough, etc. or being guided to the appropriate space for help, before and after hospitalization.

Herein, the ‘intelligent modules’ in a chatbot include user modeling modules and the natural language understanding module; which actually make the chatbot to understand the user question and choose the requisite response are based on Machine-learning algorithms. These chatbots are able to perform better by learning continuously through machine learning(ML) algorithms and human supervisors that enable the learning of the chatbot. ML techniques like reinforcement learning supervised, and unsupervised techniques are being leveraged by developers to ensure the AI chatbot becomes a good learner and thus, provide the appropriate and exact response. These chatbots have been advising patients on testing, hospitalization requirements, etc. when they receive their symptom histories.

Conclusion:

The situation that we are all engulfed in is more than overwhelming. People are dying all around, and so are hopes and dreams. But, ‘survival’ is the biggest human instinct. Machine Learning is helping in making things easier by customizing computer processors to understand the mayhem of data within this situation to try and come up with viable solutions. Our processing potentials are evolving with the help of Machine Learning and is increasingly being made available for functions all around. A data driven approach with higher accuracy can be very useful for a proactive response from the government and citizens. It is exactly what Machine Learning and its related Artificial Intelligence and Cloud Computing systems are trying to achieve.