Using Accelerated Data Analytics to Set U.S. Vaccination Priorities
Advanced analytics -- including real-time data processing -- will be needed for the upcoming COVID-19 vaccination program.
- By Todd Mostak
- January 8, 2021
From the earliest days of the pandemic, data has been essential to an effective response, yet case counts, testing data, hospitalization rates, provisional death counts and patient recoveries have been the most visible statistics. To truly meet the coronavirus challenge, the sources of data and their ultimate uses have had to go far deeper.
Strategic data sets allowed governments and health agencies to set public policies for sensible business practices, contact tracing, and social behavior. They've been able to track the efficacy of those policies in reducing COVID-19 spread and identify both hotspots and infection trends. Data has also helped on other fronts, such as in supply chain decisions affecting the deployment of PPE, medical equipment, and other vital material.
A Historic Task
Now, as humanity enters its next critical phase -- the biggest vaccination effort since the mass disease campaigns of the 1950s -- data will become even more critical. The COVID-19 vaccination phase will add another layer to an already complex data challenge. Emergency management agencies, healthcare organizations, industry groups, policymakers, community vaccination providers, occupational health centers, as well as private and commercial partners, will all need coordinated data to ensure timely, ethical vaccine administration.
As much as the vaccines themselves, information will drive the next pandemic phase at every step. A multitude of factors affect vaccine deployment, handling, and administration. Authorities will need to prioritize populations, inventories, and methods of distribution, and manufacturing. Complicating the process are the unique characteristics of each vaccine candidate, as two of the leading COVID-19 vaccines must be kept at sub-zero temperatures (one at -4°F, the other at -94°F). Once thawed, one vaccine can last for 14 days at normal refrigeration temperatures while the other lasts only five days. Moreover, many vaccine candidates require two doses, either three or four weeks apart, to be effective. The second dose must come from the same pharmaceutical supplier, adding to the recordkeeping burden.
Fortunately, data is as obtainable to direct and track vaccination as it has been to monitor COVID-19 spread. By combining public health and hospital data sets with open-source and commercial information, leaders will have the chance to gain insights to navigate all the difficult logistic and clinical hurdles. From prioritizing population risk to coordinating production of vials, alcohol pads, and even dry ice for storage and transport, data will be at the center of activity.
The greater challenge to data science, however, will be to derive insights from this data at the speed, scale, and level of granularity demanded by such a Herculean task. The World Health Organization (WHO) and the U.S. Centers for Disease Control and Prevention (CDC) are both working to identify and rank those in greatest need of vaccination. The CDC COVID-19 Vaccination Program Interim Playbook states that first-responders and essential non-healthcare workers will receive prioritization, followed by adults with high-risk medical factors. Other Phase I recipients will include adults over the age of 65, including those living in long-term care facilities.
Other factors will come into play as vaccination efforts move from late-stage Phase I into Phase II. Minority and tribal groups are disproportionately affected by coronavirus spread; moreover, hotspots must be continually monitored in case supplies must be diverted.
Time and Location Will be Decisive
Spatiotemporal data analysis (the study of data collected for both geography and time) will be invaluable to stay on top of unfolding events. As any logistics professional knows, locality is important -- but so is time. Real-time analysis is immensely helpful in coordinating an effective response. Because every state and locality involved in COVID-19 vaccination will have different needs and methods, analyzing information in real time will allow authorities to make more informed decisions.
Throughout this pandemic, decision making at the local level has been imperative. Although national and regional trends are important, stopping the spread of hotspots has required analysis by county, community, even by ZIP Code. As vaccines are administered, data sets of every kind, at every level of authority, and across both private and public sources, will need to be cross-utilized.
Another benefit of data integration will be the quick identification of best practices. In the U.S., the armed forces will be providing logistical support for vaccine distribution -- and although not actually transporting vaccines, the Defense Logistics Agency will be tracking every dose, guiding those supplies at risk of expiration to new locations to reduce waste and inefficiency. Suppliers and clinics will want to learn from early deployment scenarios to improve their operations.
The Importance of Speed and Scale
If data analytics platforms are fast enough, it will be possible for officials to make decisions immediately based on the latest information. Fighting a pandemic is like managing a firebreak -- the need to address vulnerable areas quickly is paramount. Matters progress from the theoretical to the practical very quickly.
Using the latest technologies to ingest, cross-filter, visualize, and interrogate rows of data by the billions, in milliseconds, is possible and vital. As the CDC playbook states, "...it is important for jurisdictions to have full situational awareness." Only the newest generation of advanced analytics, processing data in near-real time, and presented for instant boots-on-the-ground insight, will ensure the success of this immense undertaking.
About the Author
Todd Mostak is the CEO and co-founder of OmniSci, a pioneer in accelerated analytics that enables businesses to uncover important insights.