FAU Offers New Solutions for Managing Pandemics
From left, Behnaz Ghoraani, Ph.D., Xingquan "Hill" Zhu, Ph.D., Len Treviño, Ph.D. and Borivoje "Borko" Furht, Ph.D.

FAU Offers New Solutions for Managing Pandemics

Researchers Use Machine Learning and Social Science to Advance Our Understanding of Pandemics and How to Control Them

As SARS-CoV-2, the virus that causes COVID-19, spawns new variants in its race across the globe, scientists at FAU are working to better understand how it’s transmitted, how to design success into a clinical trial, and how different societies respond to a health emergency. Their research enhances global efforts to combat this pandemic and those that are sure to come.

The research includes two predictive models from The School of Engineering and Computer Science and one study from FAU’s School of Business. The first model incorporates mobility data — how much people move around — to predict local spread in the U.S. two weeks ahead of time. The second model predicts whether a clinical trial will succeed or fail. The study, which was a collaboration between three universities, analyzes cross-cultural data to show that countries’ responses to the pandemic are largely shaped by cultural dynamics and how they interact with more, or less, stringent governmental policies over time.

Model Uses Mobility Data to Predict Disease Spread Two Weeks Ahead

Early detection of a COVID-19 outbreak is important to saving people’s lives and restarting the economy. Since outbreaks depend to a large degree on people’s social behavior, a university-industry collaboration teamed scientists from FAU’s Department of Computer and Electrical Engineering and Computer Science with researchers from LexisNexis, a global data analytics company. The team developed a deep learning model combining data on people’s mobility — how much they move around by walking, cars, or buses — with statistics on previous COVID-19 outbreaks, government policies and demographic information such as age. The results, recently published in Journal of Big Data, accurately predict county-level outbreaks in the U.S. two weeks ahead.

The findings show that when people move around, the virus reproduces more. Because an increase in mobility increases interactions between people, especially in areas with high population density, adding mobility data to forecasting models helps scientists estimate COVID-19 growth and evaluate the effectiveness of policies such as mask mandates, according to Behnaz Ghoraani, Ph.D., senior author of the publication, associate professor in the department of electrical engineering and computer science in the College of Engineering and Computer Science, and fellow of the FAU Institute for Sensing and Embedded Network Systems Engineering (I-SENSE).

The study also demonstrates that the age of a population affects the spread of the virus: average daily cases decrease in populations that have more retirees (65 and older) and increase with populations that have more young people (14 to 44).

Using this model to predict an outbreak of COVID-19 two weeks in advance helps manage this pandemic, as well as future pandemics, by ensuring that healthcare facilities are well prepared for what’s to come.

Computational Model Predicts Whether a Clinical Trial Will Succeed or Fail

Clinical trials are necessary for determining the safety and efficacy of a drug, vaccine or device, but they are costly and time consuming, according to researchers in the College of Engineering and Computer Science. The scientists developed a computational model that can predict whether a clinical trial will succeed or be withdrawn due to a variety of factors, including insufficient enrollment, lack of funding/resources and business/sponsor decisions. The model can help clinicians design trials that optimize their efforts and reduce resources.

The researchers built a testbed of 4,441 COVID-19 trials from ClinicalTrial.gov with the objective of creating features that characterize successful COVID-19 clinical trial reports. Four types of features (statistics, keywords, drugs and embedding) were formulated to represent each clinical trial. The features characterize each clinical trial by considering clinical trial administration, eligibility criteria, clinical study design (e.g., whether placebo groups are involved), drug types, study keywords, as well as embedding features commonly used in state-of-the-art machine learning. Results show that keyword features are most informative for COVID-19 trial prediction, followed by drug features, statistics features, and embedding features. The study was recently published in PLOS ONE.

Xingquan "Hill" Zhu, Ph.D., senior author and a professor in the College of Engineering and Computer Science, points out that clinical trials involve a great deal of resources and time, including planning and recruiting human subjects. Being able to predict the likelihood of whether a trial might be terminated or not down the road will help stakeholders better manage their resources and procedures. Such computational approaches may even help our society as a whole redirect our efforts toward combating the global COVID-19 pandemic.

Cultural Norms Influence Government Reactions to COVID-19

Some nations have been more effective than others at curbing the spread of COVID-19 within their borders. To understand how some countries successfully protect the lives of their citizens during a pandemic, researchers at FAU’s School of Business, in collaboration with colleagues at other universities, combined insights from cross-cultural research, social psychology and public health literature to explain the factors that impact a nation’s response to an epidemic.

The study analyzed COVID-19 case data for the first three months of infection in 107 nations, along with factors such as stringency of government policies and how quickly they were implemented. The study was recently published in the Journal of International Business Studies.

The researchers categorized countries according to whether they prioritize individual or collective behavior; the extent to which they accept unequal distribution of power, status, and authority (power distance); the extent to which they feel threatened by ambiguities (uncertainty avoidance); and whether they prioritize ego goals versus social goals (masculinity-femininity). The study showed that stringent government policies applied early on attenuated pandemic growth, and that this effect was more pronounced in collectivistic than individualistic nations, and in high rather than low power distance nations.

Len Treviño, Ph.D., director of international business programs in the College of Business, believes that countries should provide clear messages early on regarding the potential mortality and morbidity of a pandemic, and that in order to have the intended impact, the focus and content of these messages should be tailored to the nation’s cultural context. Nevertheless, Treviño and his coauthors argue that all countries should attempt to unify their citizens by pointing out that adhering to pandemic control measures and a healthy economy go hand in hand.

Using Big Data Analytics to Model and Track the Spread of COVID-19

Public health efforts to stop pandemics like COVID-19 depend heavily on predicting how they spread. Borivoje “Borko” Furht Ph.D., a professor at Florida Atlantic University’s department of electrical Engineering and computer science and director of the National Science Foundation (NSF) Research Center (CAKE) in FAU’s College of Engineering and Computer Science, and his team collaborated with data analytics company LexisNexis to create a mathematical algorithm that models and tracks the spread of COVID-19 from the county to the global level.

Using big data about personal relationships, health center locations and known mechanisms for spread of the disease, the model looks forward, predicting infection from a given patient into the community, and backward, tracing verified infections to a possible patient “zero.”

“Let’s say someone gets COVID. In a matter of twenty minutes, we can see the whole network around this person—neighbors, friends, employees—so we can immediately start tracking who they may be in contact with. Our mathematical model immediately calculates how far anyone is from this infected person and the possibility that they might get COVID,” Furht said, who is basing the COVID-19 algorithm on a previous successful model of Ebola. The Global COVID-19 Spread Tracking project has been released as a publicly available open-source website.

If you would like more information, please contact us at dorcommunications@kome-shibahara.com.