An Introduction to CovIdentify

Karnika Singh

The COVID-19 crisis has served as a reminder that early detection of infectious diseases can not only save countless lives, but also help minimize the negative effects such pandemics have on all aspects of our society including our economy and healthcare systems. In an increasingly connected and globalized world where many people cross state and national borders regularly, there are dangers associated with undiagnosed infectious disease cases, not only for today's COVID-19 crisis, but also for future disease outbreaks. Timely detection of individuals who might be at risk of infection and need to be tested can help provide critical information for a rapid, coordinated, and large-scale response to combat this pandemic.

Wearable devices have completely changed how we track our activity and sleep. There are currently 60.5 million people in the US using 117 million wearable devices1. These devices have given us the ability to take active control of our fitness and health in a continuous and personalized way. Today, wearable devices can be used to help fight the biggest pandemic the world has seen in roughly a hundred years.

How Can Wearables Help?

The BIG IDEAS Lab at Duke University, led by Dr. Jessilyn Dunn, investigates the potential of wearable device data for digital biomarker development. We have developed the Digital Biomarker Discovery Pipeline, an open source software platform for the development of digital biomarkers using mHealth and wearables. Digital biomarkers are indicators of health outcomes developed from digitally collected data (e.g. heart rate from a vital sign monitor). Based on the increasing prevalence of smart watches with lower cost and improved functionality, we recently developed digital biomarkers to detect infection and insulin sensitivity2. Measurements from smartwatches like skin temperature, heart rate and physical activity were used to develop algorithms to detect early infection. Our team is now working to use smartwatches and smartphones for timely detection of COVID-19 infection with CovIdentify.

CovIdentify: Using Smartphones and Smartwatches Data for COVID-19 Detection?

We are working to develop CovIdentify to be an automated and personalized platform to detect infection at an early stage prior to obvious symptoms and to predict when symptoms are likely to worsen. The initiative brings together an interdisciplinary team, including Duke University doctors, scientists and engineers, to investigate if the data from your smartphone and smartwatch can help determine whether or not you have a COVID-19 infection, and how severe the infection is expected to get. The study aims to determine CovIdentify's utility to detect who should be quarantined and tested for COVID-19, and when. This can also help determine who will need specialized medical resources.

In a quote for a Medium article, Dr. Dunn explained, "If we could track a lot of people using these devices, we would start to learn where hotspots are and start to contain it before it spreads."

The study is aimed at tackling the current crisis from multiple perspectives. Tracking who has an infection and when can help determine hotspots of the infection. CovIdentify may also help extend and leverage the utility of smartwatches for tracking infection risk for not just the current pandemic but also for future disease outbreaks. The tool could be used as a personal infection tracker and by alerting people about a possible infection, could facilitate timely measures and containment. One of the most crucial elements of this study is collaborating with companies to get these wearable devices into underserved communities and onto essential workers who might not always have the means to purchase them. We have partnered with Garmin and Fitbit who have offered devices at a discounted price for the study participants to address health disparities. Since its launch on April 2, 2020, the study has enrolled over 4000 participants with support from Duke MEDx and the Duke School of Medicine Clinical & Translational Science Institute.

How Can You Contribute?

Our team uses tools from statistics and machine learning to detect periods of inflammation from wearable device measurements. With larger amounts of data, the algorithms can be more robustly trained and will lead to better predictions. Your participation in the study can help us obtain that data. By joining this study, you will play a pivotal role in helping our team learn how to detect COVID-19 early in order to facilitate better outcomes for all. Even those who do not display symptoms can spread the disease. We need people like you to help us build powerful detection models that could save lives.

Participation is easy and incurs no additional cost. You do not need to be COVID positive or own a wearable device to enroll. Anyone who owns a smartwatch or a smartphone can enroll in the study. All you need to do is answer two simple questions everyday via text or email: "Were you near family/friends/coworkers?" "Do you feel sick?" If you feel sick, we will ask you a few more questions. You can connect your Fitbit or Garmin today. We are currently working to expand to include other smart watches and will re-contact you when these become available.

You can sign up for the study here. More information can be found on the at official CovIdentify Facebook page where we also try to answer all your queries regarding the study. The study participant statistics are updated daily on the study website where you can see what your data is contributing to. You can also reach out to us at in case of any questions or concerns. Follow the BIG IDEAS Lab on twitter to stay updated on our developments. Stay safe!

1. eMarketer. Published 2018.
2. Li X, Dunn J, Salins D, et al. Digital Health: Tracking Physiomes and Activity Using Wearable Biosensors Reveals Useful Health-Related Information. PLoS biology. 2017;15(1):e2001402

This work is not peer-reviewed.