How Data is Being Used to Drive COVID-19 Response: A Q&A with Rouzbeh Razavi, Ph.D.
Rouzbeh Razavi, Ph.D., is director of the Kent State Master of Science in Business Analytics program and assistant professor in the Department of Management and Information Systems. Prior to joining Kent State University, he was Director of Group Decision Sciences at the Commonwealth Bank of Australia in New York where he led a team of data scientists and quantitative modelers to improve risk models across the bank. Before that, he was a senior scientist in the Advanced Analytics Department at SAP, where he led a team of software developers to design SAP Predictive Analytics software suite. Before joining SAP, Rouzbeh served as a research scientist at Bell Labs for several years where he received the Bell Labs’ Golden Pen Award in 2012. He has published more than 70 papers in scholarly journals and conferences and was the recipient of two best paper awards. Razavi is also an inventor/ co-inventor of 30 patent applications in the U.S., Europe and Asia.
In your previous role as a senior data scientist at SAP, you worked on a project with the Ministry of Health in Nigeria to predict the support needed for the fight against Ebola in West Africa. What did you learn through this research about the spread of disease?
In that role, I helped develop models to predict how Ebola was spreading in West Africa using a range of innovative big data analytics techniques and tools. SAP provided an application that was used by healthcare workers to help diagnose patients and use data to predict whether patients were infected with Ebola based on clusters of symptoms reported.
The ultimate objective of the project was to model the spread pattern of the virus in West Africa and to optimize the allocation of limited mobile medical resources on the ground to respond to the virus spread.
Explain how using predictive research can save lives when it comes to a health crisis such as Ebola or the current COVID-19 pandemic.
I believe that a modern threat such as COVID-19 requires utilizing modern approaches to combat the threat. There are several ways in which big data and analytics can help in fighting epidemic crises. For example, the accelerated spread of COVID-19, which has exhausted hospital capacities in many parts of the world, requires solutions that can quickly identify positive cases that are likely to escalate to critical conditions.
Also, while a significant number of positive cases experience non-life-threatening symptoms, those who suffer more severe symptoms may need oxygen and prolonged ventilators, which are also in high demand. Laboratory test results combined with demographic and radiological findings are currently being explored by researchers to assess the feasibility of integrating these data into AI-enabled decision support systems to identify critical cases as early as possible.
Another example is genomic research, which is critical to deciphering the virus genome, where big data and machine learning algorithms are deployed extensively. Similarly, bioinformatics tools are now envisioned to be critically important for vaccine development. Finally, short-term prediction models can be deployed to help emergency departments estimate the number of admitted patients with intensive care requirements, so that they can plan and prepare.
What do you see as some of the challenges facing data scientists as they work to make predictions related to COVID-19?
Data consistency and quality assurance are major issues for many data professionals who are trying to join forces and make an impact. Currently, data is coming from very diverse sources with varying levels of quality and reliability, which requires data collection and data monitoring efforts to be coordinated by international organizations such as the World Health Organization (WHO).
How do you see the response playing out differently with today’s technology compared to previous pandemics throughout history (i.e. the Spanish flu of 1918).
Technology has helped us to respond to the COVID-19 situation much more effectively compared to the Spanish Flu, which infected about a quarter of the world's population at the time. For example, in the case of COVID-19, the genetic sequences of the virus were released to the public with remarkable speed, so that countries could create diagnostic tests as quickly as possible. The possibility of exchanging information in real-time is also a significant enabler for coordinating efforts to track and combat the epidemic.
Also, with the increased rate of internet access globally, social networking platforms have become key tools to raise awareness among the public and to promote safety tips which have been shown to be very effective in slowing down the spread of the virus.
In her daily press briefings, Dr. Amy Acton, Director of the Ohio Department of Health, talks in detail about the data scientists who are compiling data related to COVID-19 cases in Ohio. What can Ohioans take from this information and the dashboard provided on coronavirus.ohio.gov?
In difficult times like this, having access to a reliable source of information that can provide fellow Ohioans with the most updated information about epidemic statistics, trends and forecasts in an easy to understand fashion, is very helpful.
Q8. As the saying goes, information is power. How will increased data help to drive ongoing response to the COVID-19 pandemic?
Data is the fuel of machine learning algorithms. Without high-quality data, the impact of the tools would be limited, if not misleading. I believe ideas such as creating large-scale COVID-19 Real World Evidence (RWE) studies that collect and aggregate relevant data from a variety of sources across the globe is going to facilitate the research and development of solutions to combat the epidemic. Organizations such as WHO will have a central role in orchestrating efforts towards creating such systems.