New MIT machine learning model shows relaxing lockdowns will spike COVID-19 cases

MIT has developed a new model of the spread of COVID-19 infection, based on publicly available data, combined with established epidemiological equations about outbreaks, and neural network-based inference. The model, described in a new report, proves accurate when trained on data spanning late January to early March in terms of anticipating the actual spread leading up to April 1 in different regions around the world, and it indicates that any immediate or near-term relaxation or reversal of quarantine measures currently in place would lead to an “exponential explosion” in the number of infections.

Researchers at MIT sought to develop a model based just on COVID-19 data, whereas others have used SARS or MERS information to inform their charting of the outbreak’s progress. Combining available COVID-19 info with a neural network-based estimation of the number of infected individuals who are under effective quarantine, and therefore no longer a likely risk of infection to others, allows theirs to go beyond existing models in terms of accurately modelling and predicting the effect of social distancing and isolation measures – and the impact should those measures be curtailed or withdrawn.

MIT’s model shows that the current infection plateau for COVID-19 in the U.S. and Italy will both take place sometime in the next week or so, which matches existing predictions available. That sounds like promising news, and it is in terms of the number of infected patients, and the impact on the healthcare system, but it absolutely should not be interpreted as meaning that this is when it’s okay to start relaxing the measures in place.

In fact, the study concludes that by “relaxing quarantine measures too soon, we have predicted that the consequences would be far more catastrophic,” according to model developer and MIT mechanical engineering professor George Barbastathis, when compared to a similar second-wave resurgence that occurred in Singapore after it began relaxing its own measures too early.