Partial Unlock for COVID-19-Like Epidemics Can Save 1-3 Million Lives Worldwide (preprint)

Background: A large percentage of deaths in an epidemic or pandemic can be due to overshoot of population (herd) immunity, either from the initial peak or from planned or unplanned exit from lockdown or social distancing conditions. Objectives: We study partial unlock or reopening interaction with seasonal effects in a managed epidemic to quantify overshoot effects on small and large unlock steps and discover robust strategies for reducing overshoot. Methods: We simulate partial unlock of social distancing for epidemics over a range of replication factor, immunity duration and seasonality factor for strategies targeting immunity thresholds using overshoot optimization. Results: Seasonality change must be taken into account as one of the steps in an easing sequence, and a two step unlock, including seasonal effects, minimizes overshoot and deaths. It may cause undershoot, which causes rebounds and assists survival of the pathogen. Conclusions: Partial easing levels, even low levels for economic relief while waiting on a vaccine, have population immunity thresholds based on the reduced replication rates and may experience overshoot as well. We further find a two step strategy remains highly sensitive to variations in case ratio, replication factor, seasonality and timing. We demonstrate a three or more step strategy is more robust, and conclude that the best possible approach minimizes deaths under a range of likely actual conditions which include public response.

System Engineering and Overshoot Damping for Epidemics Such as COVID-19 (preprint)

The goal of this paper is to contribute the perspective of a systems engineer to the effort to fight pandemics. The availability of low latency case data and effectiveness of social distancing suggest there is sufficient control for successful smoothing and targeting almost any desired level of low or high cases and immunity. This control proceeds from spontaneous public reaction to caseloads and news as well as government mediated recommendations and orders. We simulate multi-step and intermittent-with-feedback partial unlock of social distancing for rapidly-spreading moderate-mortality epidemics and pandemics similar to COVID-19. Optimized scenarios reduce total cases and therefore deaths typically 8% and up to 30% by controlling overshoot as groups cross the herd immunity threshold, or lower thresholds to manage medical resources and provide economic relief. We analyze overshoot and provide guidance on how to damp it. However, we find overshoot damping, whether from expert planning or natural public self-isolation, increases the likelihood of transition to an endemic disease. An SIR model is used to evaluate scenarios that are intended to function over a wide variety of parameters. The end result is not a case trajectory prediction, but a prediction of which strategies produce near-optimal results over a wide range of epidemiological and social parameters. Overshoot damping perversely increases the chance a pathogen will transition to an endemic disease, so we briefly describe the undershoot conditions that promote transition to endemic status.

Partial unlock caseload management for COVID-19 can save 1-2 million lives worldwide (preprint)

BackgroundA large percentage of deaths in an epidemic or pandemic can be due to overshoot of population (herd) immunity, either from the initial peak or from planned or unplanned exit from lockdown or social distancing conditions. ObjectivesWe study partial unlock or reopening interaction with seasonal effects in a managed epidemic to quantify overshoot effects on small and large unlock steps and discover robust strategies for reducing overshoot. MethodsWe simulate partial unlock of social distancing for epidemics over a range of replication factor, immunity duration and seasonality factor for strategies targeting immunity thresholds using overshoot optimization. ResultsSeasonality change must be taken into account as one of the steps in an easing sequence, and a two step unlock, including seasonal effects, minimizes overshoot and deaths. It may cause undershoot, which causes rebounds and assists survival of the pathogen. ConclusionsPartial easing levels, even low levels for economic relief while waiting on a vaccine, have population immunity thresholds based on the reduced replication rates and may experience overshoot as well. We further find a two step strategy remains highly sensitive to variations in case ratio, replication factor, seasonality and timing. We demonstrate a three or more step strategy is more robust, and conclude that the best possible approach minimizes deaths under a range of likely actual conditions which include public response.

Partial unlock model for COVID-19 or similar pandemic averts medical and economic disaster (preprint)

Data as of March 29, 2020 show that the flattening strategy for COVID-19 in the U.S. is working so well that a clean removal of social distancing (aka unlock) at any time in 2020 will produce a renewed catastrophe, overloading the healthcare system. Leaving the economy locked down for a long time is its own catastrophe. An SIR-type model with clear parameters suitable for public information, and both tracking and predictive capabilities which learns disease spread characteristics rapidly as policy changes, suggests that a solution to the problem is a partial unlock. Case load can be managed so as not to exceed critical resources such as ventilators, yet allow enough people to get sick that herd immunity develops and a full unlock can be achieved in as little as five weeks from beginning of implementation. The partial unlock could be for example 3 full working days per week. Given that not all areas or individuals will respond, and travel and public gatherings are still unlikely, the partial unlock might be 5 full working days per week. The model can be regionalized easily, and by expediting the resolution of the pandemic in the U.S. medical equipment and volunteers, many of them with already acquired immunity, can be made available to other countries.