Machine Learning Applications To Forecast The Impact of Quarantine During COVID-

Posted by AretoveTechnologies on July 26th, 2020

With concepts like social distancing and quarantining becoming normal to slow down the spread of COVID-19, scientists are trying to investigate how much these measures can really help. Researchers are trying to develop and implement machine learning models to quantify the impact of these measures in different parts of the world. 

What does the model developed by MIT look like?

The focus of this dataset is to interpret the effectiveness of the safety measures in halting the spread of the virus by leveraging a neural network and the data of coronavirus infections. Like many other models used to track the spread of any diseases, the base of this innovation is the SEIR model.  The MIT team has added more value in the model by integrating a neural network to find the number of individuals who are quarantined and not spreading the disease. The model has been successful in proving that in countries like South Korea where the government actively reacted in restricting in the public movement have managed to make the infection numbers come into a standstill position. 

But in countries like the US and Italy where the authorities took time in declaring lockdown, infections have found to rise almost exponentially. 

In trying to determine the characteristics of the spread of COVID-19 with machine learning, the team revealed that upon applying the quarantine measures the reproduction number can be effectively reduced. 

What can the model designed by RPI deliver?

COVID-19

Researchers at RPI are also relying on machine learning to assess the impact of social distancing but in a more granular form. By sourcing, the data from the Department of Health and Hygiene of New York, an RPI researcher named Malik Magdon-Ismail has created a model to forecast some of the elements of the pandemic. 

During the initial phase, the machine learning models showed that projections vary from one city to another and it has been a breakthrough in these distressing times. Another machine learning model developed by Malik sourced data up to April 10 from counties like Albany, Schenectady, Rensselaer, and Saratoga. The sample size was 855,000, and the model predicted that about 1490 per day are at a high risk of contracting infections till June 8 even if 50% of the people stay at home. 

If 70% of the people choose to stay at home the number of infections can drop down to 750 per day. Using this same model to smaller cities would be futile because the data points would be rarely updated.  That is why Ismail insisted on employing simple models that would not involve mathematics. 

To ensure that the output is robust and accurate, often a combination of multiple machine learning models is being utilized. This proves to be quite effective because there is a variety of models to cater to the characteristics of the data. To come to this conclusion, Ismail had tried and tested this technique on the data derived from the early stages of the pandemic in the USA, and by analyzing the effectsof that research, he has turned to develop a mix and match of machine learning models.  

 

Author Bio:

Ms. Aanchal Iyer is the Digital Marketing Manager and a Content Strategist for Aretove Technologies Pvt.Ltd. She has experience of 11+ years as a content consultant. Aanchal is actively involved in writing about the advancement of technologies such as AI , Data Science & in our everyday life.

 

 

The Secrets to Drawing Video Course
The Secrets to Drawing is a comprehensive learning experience designed to guide absolute beginners and intermediate artists through the core foundations of drawing through concise and "easy to digest" modules that include HD videos and ebooks.
Advertise Here