A more granular understanding of risk could reduce the need for widespread lockdowns during an outbreak
Techniques used in weather forecasting can be repurposed to provide individuals with a personalized assessment of their risk of exposure to COVID-19 or other viruses, according to new research published by Caltech scientists.
The technique has the potential to be more effective and less intrusive than blanket lockdowns in controlling the spread of disease, says Tapio Schneider, Theodore Y. Wu Professor of Environmental Science and Engineering; principal investigator at JPL, which Caltech manages for NASA; and lead author of a review of the new research that was published by Computational Biology PLOS June 23.
“For this pandemic, it may be too late,” says Schneider, “but it won’t be the last outbreak we face. It’s also useful for tracking other infectious diseases.”
In principle, the idea is simple: weather forecasting models ingest a lot of data – for example, measurements of wind speed and direction, temperature and humidity from local weather stations, in addition to satellite data. They use the data to assess the current state of the atmosphere, predict future weather patterns, and then repeat the cycle by mixing the predicted atmospheric state with new data. In the same way, disease risk assessment also leverages various types of available data to assess an individual’s risk of exposure or infection to disease, predict the spread of disease through a network of contacts humans using an epidemiological model, then repeat the cycle by mixing the predictions with new data. These assessments can use an institution’s surveillance test results, data from wearable sensors, self-reported symptoms and close contacts recorded by smartphones, and disease reporting dashboards from municipalities.
The research presented in Computational Biology PLOS is a proof of concept. However, its end result would be a smart phone app that provides an individual with a frequently updated numerical rating (i.e. a percentage) that reflects their likelihood of having been exposed to or infected with a particular infectious agent. , such as COVID-19. 19.
Such an app would be similar to existing COVID-19 exposure notification apps but more sophisticated and efficient in its use of data, according to Schneider and colleagues. These applications provide a binary assessment of exposure (“yes, you have been exposed” or, in the absence of exposure, radio silence); the new application described in the study would provide a more nuanced understanding of the ever-changing exposure and infection risks as individuals become closer to others and infection data spreads through a network of online contacts. constant evolution.
The idea was born at the start of the COVID-19 pandemic, when colleagues and partners Schneider and Chiara Daraio, G. Bradford Jones Professor of Mechanical Engineering and Applied Physics and a researcher at the Heritage Medical Research Institute, are suddenly found isolated at home and wondering how to use their scientific and technical expertise to help the world face this new threat.
One of Daraio’s pre-pandemic research goals was the development of low-cost body temperature trackers. And that raised the question: would the widespread use of these trackers allow for better tracking and understanding of the spread of COVID-19?
“We envisioned something like a weather forecast app, leveraging sensor information, infection data and proximity tracking, that people could use to adjust their behavior to mitigate individual risk,” says Daraio, co-author of Computational Biology PLOS paper.
Schneider is a climatologist who leads the Climate Modeling Alliance (CliMA), which is using recent advances in computer science and data to develop an entirely new climate model. He contacted Jeffrey Shaman, a longtime acquaintance at Columbia University. Shaman’s research on how climate change affects the spread of infectious diseases led Shaman to become interested in epidemiology and the adaptation of similar weather forecasting methods for disease modeling at the community level.
“Over the past decade, the field of infectious disease modeling, and prediction in particular, has exploded. Many approaches to disease prediction take advantage of ensemble and inference methods commonly used in weather forecasting,” says Shaman, co-author of the book. Computational Biology PLOS paper.
The team had two major challenges: to adapt weather forecasting methods for this purpose and to develop a realistic test bed to assess how well it worked.
“Conceptually, this is a very attractive idea, because weather forecasting methods have been so good at predicting the chaotic atmosphere, a notoriously difficult task,” says Caltech researcher Oliver Dunbar. “But there is no direct translation. An epidemic forecasting application has very little data to work with and only on a partial population of users. latest smart device technologies and a mathematical viral spread mockup.”
To test it, the team turned to Lucas Böttcher from the Frankfurt School of Finance and Management in Germany. Böttcher built a computer model of an imaginary city – a scaled-down, idealized version of New York – with 100,000 “nodes”, or fictitious people, then studied how well adapted weather forecasting methods predicted the spread of a disease across the population.
The results were encouraging: in simulations, the model identified up to twice as many potential exposures as would be detected by traditional contact tracing or exposure notification apps when both use the same data. .
“The methods developed in our study are relevant not only in the context of infectious disease management, but they also open new ways to combine observational data with high-dimensional mechanistic models from computational biology,” says Böttcher, co-author of the study Computational Biology PLOS paper.
Despite these promising results, real-world implementation of this technology requires appropriate levels of smart device users and effective testing campaigns to operate risk assessment software to manage and control outbreaks. If approximately 75% of a given population provide relevant information (for example, whether they have tested positive for a disease) and self-isolate when they may have been exposed, risk assessment software is sufficiently accurate in managing and controlling the COVID outbreak across the entire population. And yet, as COVID-19 vaccination rates show, buy-in from such a large part of the population is hard to come by.
Nevertheless, a promising scenario is deployment by smaller community user bases – for example, the population of a college campus – which can easily provide the software with more data than is needed to provide accurate assessments. risks that will locally reduce the spread of the disease.
“The challenge in making this system a reality is managing privacy issues, for example, the transfer of data on close contacts to a central data processing facility,” Schneider said. “That being said, only anonymized information is needed. Location information is already routinely collected for commercial purposes, and we are considering ways to harden the system against exploitation by bad actors.”
Other co-authors of Computational Biology PLOS the article features Caltech researcher Jinlong Wu and graduate student Dmitry Burov as well as former Caltech postdoc Alfredo Garbuno-Iñigo of the Instituto Tecnológico Autónomo de México; Gregory Wagner and Raffaele Ferrari of MIT (all CliMA members); and Sen Pei of Columbia University. This research was supported by Eric and Wendy Schmidt and Schmidt Futures; the Swiss National Science Foundation; national institutes of health; the Army Research Office; the National Science Foundation; the National Institute of Allergy and Infectious Diseases; and the Morris-Singer Foundation.