COVID-19 Machine Learning Uncategorized

How to Fight COVID-19 With Machine Learning

How to Fight COVID-19 With Machine Learning
banner

World pandemics are, no doubt, a dangerous threat. Coronavirus is not the first, and it won’t be the last pandemic.

More than before, we are gathering and distributing what we know about the virus. Lots of research teams around the globe are combining their efforts to collect data and create solutions.

AI and machine learning can make use of data to make objective and informed recommendations and suggestions to ensure that scarce resources are distributed as effectively as possible. Doing this will help secure lives and also assist in decreasing the load on healthcare systems and experts.

How Machine Learning Can Help In Fighting The Battle Against COVID-19 

1. Identifying who is at Risk from COVID-19

Machine learning has turned out to be of great value in predicting risks in many fields. With medical risk, particularly, machine learning is powerfully interesting in three significant ways:

Infection Risk, Severity Risk, and Outcome Risk

Machine learning can actively assist in predicting all three threats. Though it’s too soon for many COVID-19-specific machine learning research to be conducted and published, early experiments are inspiring.

Moreover, we can see how machine learning is being utilized in related areas and imagine how it could assist with risk prediction for COVID-19.

– Predicting the Risk of Infection

The early statistics revealed that essential risk factors that determine the probability of an individual is to contract COVID-19 include: Social habits, General hygiene habits, Pre-existing conditions, Age, Number of human interactions, Location, and climate, Socio-economic status and Frequency of communications,

Risk research for the present pandemic is still in the early stages. For instance, DeCapprio et al. have used machine learning to develop an initial Vulnerability Index for COVID-19. Prevention measures like wearing masks, washing hands, and social distancing are probably also to control the entire risk.

As improved data becomes available and presently, ongoing studies yield results, we will likely experience more practical applications of machine learning for predicting infection risk.

–  Predicting who is at more risk 

Immediately a person or group has been infected, and we have to predict the risk of that person or group advancing complications or requiring improved medical care. Most people experience mild symptoms, while others experience severe lung disease or acute respiratory distress syndrome (ARDS), which is very deadly.

It is not certain to treat and carefully monitor people with mild symptoms, but it’s better to begin treatment early if more severe symptoms are probably going to develop.

– Predicting Treatment Outcomes

Doctors will treat patients more effectively if it’s possible to predict the result of specific treatment methods. Using machine learning to customize treatment plans is not specific to COVID-19, and machine learning has formerly been used to predict treatment results for patients with epilepsy.

Also, researchers have used machine learning to predict responses to cancer immunotherapy. 

Because treatment options for COVID-19 are still developing, it will probably be some time before we see machine learning used for predicting results for certain treatments. But result prediction remains an essential part of risk assessment, working hand-in-hand with the infection and severity predictions.

2. Screening Patients and Diagnosing COVID-19

When a new pandemic strikes, diagnosing individuals is difficult. Testing on a massive scale is challenging. Tests are probably going to be costly, particularly in the beginning, rather than collecting medical samples from every patient and waiting for slow, expensive lab reports to be released.

 A simpler, quicker, and cheaper test (even if it’s less correct) would be beneficial in collecting data on a massive scale. This data can be used for more research, and also for screening and triaging patients. 

–  Using Wearable Technology to Screen for Resting Heart Rate

Apple made waves when they used their Apple Watch to discover common heart problems with machine learning assistance. But models in resting heart rate can be indicative of more certain issues too, and some initial research using Fitbit data points out that changes in resting heart rate can help discover or influenza-like illness patients.

This is a lengthy way from diagnosing COVID-19, mainly, but the research is still tender. 

Likewise, research from OURA, asleep, and activity tracking ring utilizes body temperature, heart rate, and breathing rate to try to discover patterns of onset, progression, and recovery for COVID-19.

– Screening Patients using Face Scans

Though, there are small accurate details available, a hospital in Florida was one of the earliest to attract attention for utilizing machine learning to assist in responding to COVID-19. Upon entering the hospital, patients are provided with an automatic face scan, which uses machine learning to detect whether they have a fever or not.

Personally, this data is likely not excessively beneficial, but when dealing with hundreds or even thousands of patients, all piece of data is essential in assisting triage them efficiently.

3. Speeding up Drug Development

In response to the recent pandemic, it’s essential to develop a vaccine, a suitable diagnostic method, and a drug for treatment, quick. Current methods involve lots of trial and error, which takes time. It can take months to isolate even one viable vaccine candidate. 

Machine learning can quicken this process, importantly, without sacrificing quality control. When researchers were trying to discover small molecule inhibitors of the Ebola virus, they found out that training Bayesian ML models with viral pseudotype entry assay and the Ebola virus replication assay data helped speed up the scoring process. (Scoring involves assigning every molecule a value based on how possible it is to help.)

This fastens process quickly identified three potential molecules for testing.

In situations such as the COVID-19 pandemic, where a virus is spreading speedily, getting more correct scores faster is essential to speeding up drug development.

4. Identifying Effective Existing Drugs 

Companies spend massive time and money getting new drugs approved. They need to be as specific as they possibly can that these drugs won’t have any dangerous side effects.

This process saves us, but it also slows us down during a pandemic, when we need a quicker response.

One way is to repurpose drugs that have been tested and used to treat other diseases. But there are lots of drug candidates, and there is no time to check them all, so how do we discover the accurate one?

Machine learning can help us prioritize drug candidates rapidly by automatically developing knowledge graphs and Predicting interactions between drugs and viral proteins.

5. Predicting the Spread of Infectious Disease using Social Networks

Usually, amid this pandemic, the government works together with the health system to keep track of the number of infected people and places. For instance, daily or weekly. They are responsible agency counts and publicizes the number of new patients diagnosed with the disease. 

But one of the issues here is that there might be a big gap (in time and space) between getting the disease, developing the initial symptoms, and testing positive.

6. Understanding Geographic Distribution of Cases relative to the Economic Contribution

Clustering algorithms can assist in tracking this indicator. It enables grouping a set of objects in such a way that objects in the same group (called a cluster) are more alike to each other than to those in other groups (clusters). 

For instance, Maharashtra State contributes a reasonable portion to the Indian entertainment industry and economy, which has the most significant number of COVID-19 cases.

According to the India Brand Equity Foundation, the Indian media and entertainment industry is expected to attain around Rs 307,000 crore (US$ 43.93 billion) by 2024, which may cause an adverse hit in the current scenario.

7. The Extent of Behavior Shift

Sentiment analysis is used to know about the change in behavior. It is massively being used for social media monitoring, brand monitoring, the voice of the customer (VOC), customer service, and market research.

In the relation of COVID19 lockdown, it is very vital to analyze the post COVID19 period on the behavior shift in spending on socializing like eating out at restaurants, entertainment, etc.

8. Understanding Viruses through Proteins

To know about viruses such as COVID-19 is to understand its proteins, whether and how we get sick depends on how these proteins interact with our bodies. But interpreting them is challenging.

The following use cases give examples of how machine learning can help enhance our understanding of viruses by analyzing their proteins: Predicting viral host protein-protein interactions and predicting protein folding.

9. Planning to Attack the Virus

Compared to normal traditional vaccines, which have non-functioning pathogens, epitope-based vaccines are more secure they prevent disease without the risk of potentially dangerous side effects. 

Identifying accurate epitopes can be time-consuming and costly. With the present COVID-19, locating epitopes rapidly speeds up the process of creating effective vaccines.

This is where machine learning can assist. Support vector machines (SVM), hidden Markov models, and artificial neural networks (particularly deep learning) have all proved to be quicker and more correct at identifying epitopes than human researchers are.

10. Figuring out the Hosts in the World

A zoonotic pandemic like the COVID-19 pandemic is caused by an infectious disease that begins in a different species (such as bats) and then spreads to humans. Viruses like Ebola, HIV, or COVID-19 can survive unnoticed in the natural world for a while, waiting for the next mutation and the next chance to infect us.

They hide in animals called reservoir hosts that are not affected by the illness. 

Thanks to the massive advances in technology, Whole-Genome Sequencing (WGS, the process of determining an organism’s complete DNA sequence) has become cheap and quick.

Research has revealed that machine learning models can use genome sequencing data alongside professional knowledge to identify the species that probably acted as hosts for the disease. 

By looking at a little subset of species, we can dramatically quicken discovering these pathogens in the wild.

11. Predicting the Risk of the new Pandemic

Predicting whether a pandemic is coming based on basic reasoning will assist all the medical professionals on how to manage any disaster.

For example, Influenza A exists only in the creatures of a bird with a small population, but it can catch a human. Therefore the scientists working on Influenza A only allow 67,940 protein series from the database, and they are thoroughly filtered so that the database has a set of Influenza proteins.

The researchers use machine learning to identify the potentially zoonotic strains of influenza with high levels of correctness. Lots of work needs to be done to accomplish prediction models for direct transmission, but knowing which strains of influenza are probably to leap is a vital initial step in preparing for the future pandemic. 

Conclusion

Machine learning is an essential tool in fighting this present pandemic. If we can embrace the opportunity of gathering data, improve our knowledge, and combine our skills, we can save many lives both presently and in the future.

This impact sparks the use of econometric ML models offering empirical analysis to economic relationships and enabling data-driven decisions. Regardless of the models suggested, a conclusive approach may need testing with various algorithms.

The transformation in training data may need an evolutionary approach for choosing an interpretable model or model-agnostic method

banner

Related posts