Source: Gu, Y. COVID-19 projections using machine learning. https://covid19-projections.com. Accessed 27 Jan 2021.
Since the early days of the pandemic, we've heard wildly varying predictions of the number of COVID cases and deaths that we might see in the US. Back in March 2020, Dr. Anthony Fauci, Chief Medical Advisor to the President, predicted between 100,000 and 200,000 Coronavirus-related deaths in the US based on projections at that time. Ten months later, we have surpassed 425,000 deaths. As vaccines are approved, new variants emerge, changing the potential trajectory day by day.
Despite what many consider to be a poor vaccine rollout in the United States, it appears that we may have just crossed a critical benchmark in the fight against COVID-19.
Data scientist and machine learning expert Youyang Gu predicts a path to herd immunity on covid19-projections.com based on a variety of data points: current infections, past infections, social distancing measures in place in each community, number of 1st and 2nd doses of vaccines administered, number of vaccines expected to be administered (including potential new ones) with their efficacy taken into consideration, and so on. This data also considers vaccine efficacy against new variants of SARS-CoV-2.
Gu makes the following predictions:
We estimate COVID-19 herd immunity (>70% of population immune) will be reached in the US during summer 2021 (Jun-Aug 2021). At a high level, herd immunity is a concept in which a population can be protected from a virus if enough people possess immunity.
The prediction about when herd immunity will be reached has been relatively stable, changing only slightly as new factors are taken into consideration. Gu's graphs are interactive; we can see the history and the potential future updated daily as new data comes in.
At the time herd immunity is reached, roughly half of the immunity will be achieved through natural infection, and the other half will be achieved through vaccination.
As more people get vaccinated, the amount of new infections is slowly beginning to drop.
New infections may become minimal before herd immunity is reached. But due to imported cases and localized clusters, it is unlikely that new infections will drop to zero until at least 2022.
This is important. Herd immunity is a critical benchmark to hit, but cases are expected to drop significantly before herd immunity is reached.
Deaths may drop to low levels even earlier (May-Jul 2021), in part due to a vaccine distribution strategy that initially prioritizes high-risk individuals. Once deaths fall to minimal levels, we may see a relaxation of restrictions. Summarizing the above findings, our best estimate of a complete “return to normal” in the US is mid-summer 2021 (June/July 2021).
It is not yet known whether any of the available COVID-19 vaccines provide sterilizing immunity. At the moment, researchers hypothesize that Pfizer and Moderna's vaccines do NOT provide sterilizing immunity, but that they prevent serious illness and death in 94-95% of people who receive them. What does this mean? A person who is vaccinated will likely not become seriously ill, but a small percentage may still become infected asymptomatically.
We estimate roughly 70-75% of the US population (230-250 million) will receive at least one dose of the vaccine by the end of 2021, with children being the last group to receive it (fall 2021).
None of the vaccines on the market have been approved for use in children yet. While infection rates may drop significantly by the summer, children are likely to be the last ones to be vaccinated.
We estimate around 30-40% of the US population (~115 million) will have been infected by the SARS-CoV-2 virus by the end of 2021. That is an additional 50 million infections since mid-December 2020.
This is a significant number. Despite taking precautions for a year+, 30-40% of the US population will become infected anyway.
This translates to a final US COVID-19 death toll of roughly 600,000 (±100,000) reported deaths, or ~300,000 additional deaths since mid-December 2020.
Unfortunately, the projections include hundreds of thousands of additional deaths, even with vaccinations being given to over a million people per day in the United States.
For the majority of the assumptions, the exact numerical values have relatively minor impacts on the final results (e.g. changing vaccine efficacy from 85% to 90% results in a 10-day shift in when herd immunity is reached).
This is poignant. The projections indicate that even with increased vaccine efficacy, we are likely looking at a long road to herd immunity.
Gu's projections are updated regularly and the raw data is available for anyone to view.
January 28: We are releasing raw and adjusted CDC vaccination data for download on GitHub. It contains every CDC vaccination update since they started releasing data on Dec 20, 2020. January 25: In December, we reported on why the CDC overestimates true infections in the US. They have since adjusted their estimates, which now closely matches covid19-projections.com.