MIT engineers have developed a new model that taps a neural network to evaluate different quarantine efforts and predict their effect on the spread of the novel coronavirus. The model is based on actual data from the ongoing pandemic, which has spurred various levels of lockdowns in states and countries around the world. Based on the data, quarantine efforts should continue to ‘flatten the curve.’
The project kicked off with the data from Wuhan, China, later incorporating data from Italy, South Korea, and the United States, as well. The work combines a neural network with the existing SEIR model, which stands for ‘susceptible,’ ‘exposed,’ ‘infected,’ and ‘recovered.’ The team looked specifically at how quarantines in different countries impacted the rate at which people became infected with the virus.
After 500 iterations of training, the model was able to predict patterns of infection, finding that in places where ‘strong measures were implemented quickly,’ the quarantine effort could effectively curb the spread of the virus. In places like the US where the quarantine efforts were rolled out slowly, the effort to slow the spread has been less effective.
Using the data, the model predicts when the coronavirus ‘plateau’ will take place, finding that the cases aren’t likely to start stagnating until some time between April 15 and April 20. By the time this plateau starts, the model estimates that up to 600,000 people the US may have contracted the novel coronavirus, which causes a disease called COVID-19.
The researchers warn that relaxing quarantine efforts too early may prove disastrous, leading to a greater number of cases or the second wave of outbreaks. Coming alongside the new study is a statement from the World Health Organization, which warns that the ‘next few weeks will be critical for Europe.’
The organization says that restrictions on quarantine can only be made when multiple conditions are met, including when the transmission is controlled, hospitals have capacities to meet all the needs of the public, workplaces have effective preventative measures in place, and more.