What data sources are used in this new pandemic model?

The model uses data on population-weighted density, lockdowns, cell phone activity (mobility), weather patterns, age distribution, and Google Trends for face mask searches. This allows for a more nuanced understanding of transmission dynamics.

How does this pandemic model differ from standard compartmental models like SIR?

While built on similar epidemiological principles, this model's key innovation is identifying how model parameters depend on local conditions and then applying a universal model across different regions, rather than fitting separate, localized models.

Does weather affect virus spread according to this model?

Yes, the model found that virus spread is faster in both very cold and very warm weather. This is hypothesized to be a proxy for how much time people spend indoors, which increases with extreme temperatures.

What is the significance of mortality data in understanding pandemic spread?

Surprisingly, places with higher mortality rates (around 0.1-0.2% of the population dying from COVID-19) showed a significant slowdown in virus spread. This is consistent with acquired immunity within the population.

What is the herd immunity threshold for COVID-19 according to this new research?

Based on mortality data and an assumed infection fatality ratio of 1%, the model suggests herd immunity might be reached when 10-20% of the population is infected, which is lower than the traditional 60-80% estimate for homogeneous societies.

How has the epidemiology community responded to this new pandemic model?

The response has largely been one of dismissal or ignorance, according to the researcher. While some epidemiologists are exploring similar ideas, the broader community has not widely engaged with their findings.

Key Moments

A New Model for the Covid Pandemic

Sabine Hossenfelder

Science & Technology4 min read21 min video

Oct 28, 2020|35,423 views|1,437|495

epidemic pandemic covid pandemic models science of pandemic physics niayesh afshordi herd immunity science science news corona virus new model for pandemic

Save to Pod

Key Moments

TL;DR

Astrophysicist Niayesh Afshordi presents a new pandemic model using population-weighted density and other factors, differing from traditional models by identifying universal drivers.

Key Insights

A new COVID-19 model developed by astrophysicist Niayesh Afshordi focuses on population-weighted density as a key predictor of virus spread.

The model integrates various data sources, including mobility data (Google), weather patterns, and demographic information (age distribution).

Unlike traditional models that often use fixed parameters or region-specific adjustments, this model aims for universal parameters applicable across different locations.

The research suggests that weather plays a significant role, with the virus spreading more quickly in both very cold and very warm conditions, potentially due to people spending more time indoors.

Surprisingly, high mortality rates in a region correlate with a slowdown in virus spread, which is consistent with the concept of increasing population immunity.

The model's findings on herd immunity thresholds suggest they could be lower than classically estimated, particularly when accounting for factors like super-spreading events.

ORIGINS OF THE PANDEMIC MODEL

Astrophysicist Niayesh Afshordi, initially known for his work on black hole echoes, shifted his focus to pandemic modeling during the widespread lockdowns. Motivated by the scientific imperative to understand natural phenomena and the abundance of available data, Afshordi, along with a virologist colleague, began analyzing the factors driving virus transmission. This interdisciplinary approach aimed to move beyond anecdotal observations and leverage scientific methodologies to comprehend the dynamics of the COVID-19 pandemic.

KEY PREDICTIVE FACTORS IDENTIFIED

The core of Afshordi's model lies in identifying critical drivers of virus spread. While population density is a common consideration, the model emphasizes 'population-weighted population density,' which reflects how many people are in close proximity to each other, rather than an average over a large area. This metric proved to be a strong predictor of transmission speed. The model also incorporates other dynamically accessible data, such as real-time mobility information from sources like Google and historical and forecasted weather data.

DATA INTEGRATION AND MODEL CALIBRATION

The model was initially developed and calibrated using data from the United States, chosen for its relative data homogeneity. The analysis focused on the county level, fitting a consistent set of parameters to approximately 500 counties with significant epidemic activity. This approach requires fitting around ten parameters to roughly 7,000 data points, allowing for predictions based on local conditions like population density, weather, and mobility restrictions, all of which are publicly available data.

THE PHYSICS-BASED EPIDEMIOLOGICAL MODEL

Afshordi's model is fundamentally an epidemiological model that simulates the interaction of people, virus transmission, and incubation. Unlike simple fitting functions, it models the real-world process of infection spread. This allows for predictive capabilities: by constraining the model with data, it can forecast outcomes under different policy interventions, such as lockdowns, mask mandates, or changes in social behavior. The model quantifies the dependence on various quantifiable physical and social properties.

WEATHER AND MORTALITY AS SIGNIFICANT FACTORS

The model highlights surprising correlations with environmental factors. It indicates that the virus spreads more rapidly in both very cold and very warm weather, suggesting an optimal temperature range where transmission is slower, likely due to people spending more time indoors during extreme temperatures. Another significant finding is the correlation between high mortality rates and a slowdown in virus spread. This is explained by the principle of increasing population immunity as more individuals become infected and potentially recover.

DEMOGRAPHICS AND HERD IMMUNITY THRESHOLDS

Demographic data, specifically the age distribution of a population, is also integrated into the model. Older populations are found to experience faster epidemic growth, influencing predictions. Furthermore, the model's findings on herd immunity are noteworthy. By analyzing mortality data, it suggests that herd immunity might be achieved at lower infection rates (10-20% infection, corresponding to 0.1-0.2% mortality with an assumed infection fatality ratio of 1%) than classically estimated, especially when accounting for phenomena like super-spreading events, aligning with recent independent research.

THE UNIVERSAL MODEL APPROACH

A primary distinction of Afshordi's model is its pursuit of universality. While traditional epidemiological models often adapt parameters for different regions or assume them to be constant, this model seeks to identify universal laws governing epidemic spread, much like Kepler's laws for planetary motion. By objectively measuring drivers across diverse datasets (Google, weather, epidemic history), the goal is to apply the same set of parameters to epidemics everywhere, offering a consistent framework for understanding and predicting outbreaks globally.

RECEPTION BY THE EPIDEMIOLOGY COMMUNITY

The reception of this physics-heavy model by the established epidemiology community has been largely one of silence, with many researchers choosing to ignore it. While Afshordi acknowledges that domain expertise differs, he believes the model's data-driven, analytical approach warrants serious consideration. He notes that a small subset of epidemiologists, particularly those focusing on herd immunity and super-spreading, have found the work consistent with their own findings, suggesting potential for interdisciplinary collaboration despite the initial lack of engagement.

Mentioned in This Episode

●Software & Apps

●Companies

●Organizations

●People Referenced

Key Factors for Pandemic Modeling

Practical takeaways from this episode

Do This

Utilize population-weighted density for predicting virus spread.

Incorporate data on lockdowns, weather, and cell phone activity.

Use consistent parameters across different regions to find universal laws.

Consider age distribution, with older populations showing faster epidemic growth.

Monitor Google Trends for face mask searches as a proxy for usage increase.

Avoid This

Rely solely on average population density; focus on proximity.

Assume parameters are constant across all regions; they depend on local conditions.

Ignore the impact of weather; both cold and warm temperatures can increase spread.

Discount the significant effect of mortality on slowing down virus spread.

Disregard findings on lower herd immunity thresholds, especially in heterogeneous societies.

Herd Immunity Thresholds and COVID-19

Data extracted from this episode

Condition	Infected Population Percentage	Reference
Classical estimate (R0=3-4)	60-80%	Textbooks
With super-spreading events	Potentially as low as 40%	Published papers
Based on fitting data (controversial)	10-20%	Gabriela Gomez and collaborators
Based on mortality (0.1-0.2% population death)	10-20% (assuming 1% IFR)	Nia Jasof Shorty's model

Common Questions

Population-weighted density measures how many people are in close proximity to you, rather than just the average density of a region. The model suggests this is a crucial factor in predicting how quickly a virus spreads from person to person.