Exponential Growth in Epidemics: Detecting the End of COVID-19 Spread with a Log-Log Graph | MinutePhysics

Overview

In this MinutePhysics discussion, Henry Reich walks through why COVID-19 numbers change so quickly and why exponential growth is hard for people to grasp. He highlights a visualization by Atish Bhâtia that plots the growth rate against the total number of cases on logarithmic scales, revealing where exponential spread is slowing or stopping.

• Exponential growth is difficult to intuitively recognize in daily news data.
• The graph uses a log-log plot of new cases (growth rate) vs total cases to reveal straight-line trends characteristic of exponential growth.
• A weekly rate of new cases helps show slowdown more clearly than looking at total counts alone.
• Caveats include data limitations, testing changes, and reporting delays that affect interpretation.

Introduction

Henry Reich of MinutePhysics discusses why piecing together a coherent picture of the COVID-19 pandemic from daily news is challenging. He notes that exponential growth drives rapid changes in numbers, which quickly become outdated, and he emphasizes the importance of understanding trajectories rather than solely tracking current counts. The video references a global visualization by Atish Bhâtia that maps how countries move along an exponential-growth path on a shared trajectory, but with timing differences depending on interventions and testing capacity.

Three Core Ideas Behind the Visualization

The visualization rests on three key ideas. First, plotting on a logarithmic scale is natural for exponential growth and allows comparisons across countries with very different scales. Second, focusing on the growth rate, rather than the absolute number of cases, makes it easier to detect when growth is slowing. Third, plotting the growth rate against the total number of infections instead of against time captures the essence of exponential dynamics: the rate of new infections is proportional to the current pool of infected individuals. When growth is plotted as new cases versus total cases on log scales, exponential growth appears as a straight line, making deviations easier to spot.

Idea 1: The Logarithmic Scale

On a log scale, tick marks progress by factors of ten, which scales both small and large numbers so that growth looks consistent across countries. This scaling lets us compare countries with vastly different case counts without one being visually dominated by the other. Reich stresses that this perspective is essential for identifying when a country leaves the exponential growth regime.

Idea 2: Emphasizing Change Over Time

While total case counts can suggest trend, they can obscure whether the outbreak is accelerating or slowing. By examining the growth rate—the number of new cases in the last week—the visualization makes it easier to see when the curve is flattening or turning downward, signaling a potential end to the exponential phase.

Idea 3: Not Plotting Time, But Infections Versus Growth

The defining feature of exponential growth is that new cases are proportional to existing cases, a relationship that becomes a straight line when plotting new cases (growth rate) against total infections on a log-log scale. This approach abstracts away calendar time and focuses on the disease dynamics that matter for forecasting and policy planning.

Interpreting the Graph

Reich notes that the graph shows countries generally following similar trajectories, just shifted in time. Public health measures such as testing, isolation, distancing, and contact tracing begin to bend the trajectory, shown as deviations from the main sequence. The visualization also reveals that even countries with few current cases may follow a similar path if unchecked, underscoring the need for proactive measures to prevent exponential spread.

Key Caveats and Limitations

The visualization is designed to highlight deviations from exponential growth rather than to provide exact counts. There are several caveats: (1) 10,000 and 1,000 look close on a log scale, which can distort perceived seriousness; (2) the log X axis can mask resurgences after downturns; (3) time is represented through animation, not an axis, which may be unfamiliar to some viewers; (4) the data reflect detected cases, not the true number of infections, and are affected by testing capacity, reporting quality, and delays. Reich underscores that the graph is a pessimistic but informative tool for seeing trends rather than relying on raw counts alone.

Practical Takeaways

The discussion emphasizes that understanding trends and rates of change is crucial for forecasting and policy planning. By focusing on growth rates and using log-log plots, public health officials and informed citizens can better gauge whether interventions are working and whether a country is approaching the end of exponential growth. The video ends with appreciation for the collaborators and a nod to the broader goal of making trustworthy data-driven insights accessible during a pandemic.

Conclusion

The core message is that plotting growth rate versus cumulative cases on a log scale provides a powerful lens for assessing how COVID-19 spreads globally and where progress is being made. While no visualization is perfect, this approach highlights deviations from exponential growth and offers a clearer picture of where the end of the tunnel might be in sight.