Current Implicit Policies May Not Eradicate COVID-19

Ali Jadbabaie, Arnab Sarker, Devavrat ShaharXiv 2022 · DOI · arXiv

Introduction: The Endogenous Feedback Loop

Most epidemiological models, such as the classic SIR (Susceptible-Infectious-Recovered) framework, treat human behavior and government policy as "exogenous" variables—external inputs that are independent of the system's state. However, the COVID-19 pandemic demonstrated that behavior is fundamentally "endogenous." As case counts rise, populations naturally adopt protective measures like masking or social distancing; as cases fall, these behaviors relax.

This research from MIT argues that this reactive behavior forms a feedback loop that prevents the virus from being eradicated. While current "implicit policies" are sufficient to prevent runaway growth, they are mathematically destined to reach a stable, non-zero equilibrium rather than total eradication.

Modeling Behavior as a Control System

The authors propose a second-order affine dynamical system to model the growth phase of the epidemic, using a mechanical analogy of position, velocity, and acceleration.

  • The Acceleration Analogy: Rather than controlling the "velocity" (the contact rate), the model assumes people control their "acceleration"—the rate at which they adjust their social interactions.
  • Implicit Feedback Law: The population’s behavior is modeled as a feedback law where the change in social ties is a function of current case counts and the recent rate of change.
  • PD vs. PID Controllers: In control theory terms, current behavior acts as a Proportional-Derivative (PD) controller. It reacts to the current state and its trend. To achieve eradication, the authors argue the system requires an integral term—a "memory" component that reacts to the cumulative burden of the disease since the start of the outbreak.

Results: Accuracy and the Path to Eradication

The model's simplicity belies its predictive power. Using only three learned parameters, the system achieved high accuracy across all 50 US states.

Distribution of R² values for the implicit feedback model compared to CDC ensemble forecasts across US states.

As shown in the figure above, the model achieved an $R^2$ value of 0.966 (in-sample) and 0.949 (out-of-sample). Its performance is comparable to, and in 32 states superior to, the complex ensemble models utilized by the CDC. This suggests that the "acceleration" analogy captures a fundamental truth about collective social behavior during a pandemic.

The most critical insight, however, concerns the long-term trajectory of the virus. Under current PD-like behavior, the system reaches a stable point where infections persist. As cases drop, the perceived risk decreases, leading to a relaxation of control and a subsequent resurgence.

Comparison of actual weekly cases against model fits and the simulation of a proposed feedback policy for eradication.

As illustrated in the simulation above, the "Proposed Feedback Policy" (incorporating an integral term) is the only strategy that successfully drives the infection curve to zero. Without this shift in policy—moving from reacting to recent trends to accounting for the cumulative burden—the virus is likely to remain endemic.

Limitations and Future Directions

While robust, the model is intentionally parsimonious. It does not explicitly account for biological variables such as varying incubation periods, the emergence of new variants, or the effects of waning immunity. Additionally, the model's effectiveness depends on the accuracy and timeliness of case reporting; data latency can significantly degrade the performance of any feedback-based controller.

Future research should explore how to effectively communicate "cumulative costs" to the public. Encouraging "integral" behavior—where the collective response is sustained until the total burden is addressed—may be the key to moving beyond stable persistence toward true eradication.

Summary updated

Abstract

Successful predictive modeling of epidemics requires an understanding of the implicit feedback control strategies which are implemented by populations to modulate the spread of contagion. While this task of capturing endogenous behavior can be achieved through intricate modeling assumptions, we find that a population's reaction to case counts can be described through a second order affine dynamical system with linear control which fits well to the data across different regions and times throughout the COVID-19 pandemic. The model fits the data well both in and out of sample across the 50 states of the United States, with comparable $R^2$ scores to state of the art ensemble predictions. In contrast to recent models of epidemics, rather than assuming that individuals directly control the contact rate which governs the spread of disease, we assume that individuals control the rate at which they vary their number of interactions, i.e. they control the derivative of the contact rate. We propose an implicit feedback law for this control input and verify that it correlates with policies taken throughout the pandemic. A key takeaway of the dynamical model is that the "stable" point of case counts is non-zero, i.e. COVID-19 will not be eradicated under the current collection of policies and strategies, and additional policies are needed to fully eradicate it quickly. Hence, we suggest alternative implicit policies which focus on making interventions (such as vaccinations and mobility restrictions) a function of cumulative case counts, for which our results suggest a better possibility of eradicating COVID-19.
BibTeX
@misc{jadbabaie2022current,
  author = {Ali Jadbabaie and Arnab Sarker and Devavrat Shah},
  title = {{Current Implicit Policies May Not Eradicate COVID-19}},
  year = {2022},
  eprint = {2203.15916},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2203.15916},
}
Figure