The Missing Link in COVID Planning is Real-Time, Data-Driven Risk Modeling
Our national COVID-19 policy for reopening the economy is a patchwork of plans with varying mixes of masking, social distancing and testing. What’s missing, as evident in the results of these measures on the ground, is the collective action that unites the strength of government and industry with our nation’s technical prowess. We must not only implement widescale rapid testing, but link it to high-speed risk modeling to inform policy and responses at the local, regional and national levels.
We should start with smarter testing. We need to substantially increase the number of tests we conduct each day, while taking actual risk into account to deploy these limited resources effectively. For example, cities with dense housing, nursing homes and vulnerable populations should be tested more frequently than sparsely settled areas. The same is true for businesses. A store with two employees would test less often than a meat processing plant with 500. Widespread antibody testing would bolster our ability to assess community risk and therefore determine regional testing protocols.
But not all tests are created equal. To date, faster tests are less accurate individually, but if conducted in large volume are highly effective in detecting potential outbreaks in real time and prompting swift contact tracing. People who test positive on rapid tests should be offered the next level of testing — the lab-administered PCR (polymerase chain reaction) test that finds even small amounts of the virus’s genetic material — to confirm the result.
Most critically, we need to continue pushing the R&D envelope to create tests that are precise, inexpensive and easy to administer. This past spring, the National Institutes of Health (NIH) launched a $500-million Shark Tank-style competition designed to do just that — produce a variety of tests that can be quickly evaluated, validated, manufactured and distributed. Credible ideas, including saliva- and paper-based point-of-care tests were funded immediately. More recently, the agency has raised that commitment to $1.5 billion for ongoing R&D in rapid testing that is coordinated with large-scale manufacturing and distribution plans. There are more than 100 tests now being developed for validation and commercialization.
The devices we develop must be simple point-of-care tests that people can use at home or at their local pharmacy chain. The more complicated, the less reliable they will be to perform correctly and the greater the contamination risk. Furthermore, methods for registering the data and transmitting it to cloud-based regional data centers for storage, analytics and further action also should be simple, including via point-of-care testing devices with smartphone interfaces that produce a label, or identification, for each test, and potentially a location. At NJIT, chemist Omowunmi Sadik is developing a disposable, paper-based biosensor that detects the presence of the virus, as well as its concentration level, thus tracking the disease’s progression. It will be read using a smartphone powered by a small rechargeable battery.
The use of data mining, deep learning and AI tools will allow us to develop population-based models to update risk mitigation strategies. These models would incorporate past and current data on infection rates; contact-tracing information; population density in local areas and within structures such as residential housing, schools, shopping malls and corporate buildings; demographics such as age distribution, gender and ethnicity; and healthcare statistics from electronic health records and regional hospitals to find hotspots, analyze the infection spread and predict the future trajectory in local communities.
Establishing these data centers should be a priority. Housed in municipalities and connected to county, regional and state planning, they would be synchronized with national data centers and NIH resources to develop multi-level population models that show community spread. They would allow us to assess ongoing hospitalization and critical care needs, to distribute supplies where they are most needed and to generate location-specific safety protocols for community restrictions to help mitigate risk under different scenarios as dropping temperatures drive us inside. The local governments that run them would work in collaboration with relevant private industries, while universities would provide expertise in modeling research.
The pandemic has attacked our health and economy and changed the manner in which we live, operate businesses and deliver education in ways inconceivable just six months ago. But there is some gain in this pain. It has also brought us together in urgent collaborations that are ushering in new modes to safely communicate, work and protect our health. This investment in infrastructure, technology and resilience will benefit society not just in case of a new pandemic, but in other profound ways we are just beginning to imagine.
Atam Dhawan, senior vice provost for research at NJIT and chair of the NIH Point-of-Care Technologies Research Network Advisory Board
