When we look back at the COVID pandemic, what will hindsight tell us? Will we remember the turn of the decade as the year that finally brought real change to pandemic preparedness, or will our eventual return to “normal” stymie our progress?
Although epidemiologists have long warned about the potential for global pandemics, their admonitions have largely gone unheeded. However, industrialized animal farming practices, increased human-animal contact, globalization, decreasing biodiversity and other factors all point to the likelihood of another zoonotic disease (one transmitted from animals to humans) with pandemic potential .
A slim silver lining of the current COVID-19 pandemic is that it can help us better prepare for future outbreaks—if we harness what we’ve learned correctly. In particular, we can better leverage one of the most crucial resources we have when it comes to pandemic preparedness: real-world data.
The pandemic has created a trove of data that can help us plan for future disease outbreaks. The abundance of research on the U.S. pandemic response provides insight into the benefits and consequences of various courses of action, and we can leverage this knowledge for future response.
One of the main takeaways is the need for the health care system to have real-time visibility. While observers have stated again and again that the ineffective roll-out of testing was (and still is) one of the U.S.’s biggest failings in getting ahead of COVID-19, there is a wealth of other data that can offer insight into the virus’ spread. We need to improve in collecting, sharing and analyzing this real-world data so we can rapidly recognize COVID-19 symptoms, identify effective treatments and more quickly track the spread.
For example, when the pandemic began, information disseminated by public health organizations identified sore throat, shortness of breath, cough and fever as symptoms. However, months later additional symptoms like rashes and skin discoloration—such as on the toes and feet—were recognized as potential indicators of the virus. Additionally, what has been termed “silent hypoxia”—COVID-19 causing critically low blood-oxygen levels without any noticeable external effects on breathing—killed many patients before doctors knew to be on the lookout for it.
Why weren’t we able to recognize these symptoms sooner? The electronic health records (EHRs) in which physicians document patient visits do not allow for an easy, effective way for data to be shared at scale. If de-identified patient data could be mined at a national level, artificial intelligence and machine-learning algorithms could have identified patterns far faster than it took isolated researchers working with small patient pools. Instead of examining COVID-19 data holistically, within six months, researchers had published over 23,500 papers—a wealth of information, but too much data for any one individual to possibly parse through and identify the valuable studies.
Centralizing data access could have not only sped the identification of COVID symptoms but also allowed for rapid studies of effective treatments. Researchers could use a truly robust database to analyze and identify which treatments are most effective for patients with various underlying conditions or disease histories.
Furthermore, using machine-learning techniques within a shared database could generate predictive insights, showing the patterns in communities that precede outbreaks and helping dictate where and when lockdowns and social distancing orders should be implemented. Several countries are already using unconventional data sources, like de-identified cell phone and fitness tracking data, to predict COVID outbreaks.
