As governments and communities fight to slow the spread of the novel SARS-CoV-2 virus and related COVID-19 disease, necessary responses such as mandatory social distancing and business closures have strained industries and economies. The sudden disruption of interconnected global supply chains is affecting everything from the fabrication of critical hospital ventilators and surgical masks to the shipping of bottled water, pet food, and toilet paper to retailers and consumers around the world. 

Had we been better able to predict interruptions in production and consumption—or even that consumers would hoard pasta and toilet paper—we might have devised a more adaptive supply-chain system that could continue to meet demands in the face of crises. 

In an era of ubiquitous AI and analytics tools, better predictive analysis of business systems and human behaviors should improve our chances of building resilient economies that aren’t so easily thwarted, even in the face of a global pandemic. We have all the pieces in place to make such predictions; we just need to combine them in novel ways that will allow us to perform the necessary analyses rapidly and at scale. 

In an era of ubiquitous AI and analytics tools, better predictive analysis of business systems and human behaviors should improve our chances of building resilient economies that aren’t so easily thwarted, even in the face of a global pandemic.

Consumer Signals: Existing efforts to model consumer behaviors give us useful predictive signals. Building on work in retail demand forecasting, price elasticity, and customer segmentation, data scientists should create new models to predict how soon, based on selected conditions, various key indicators are likely to return to their pre-COVID levels. For example, in the airline industry, travel for in-person meetings might be significantly reduced as businesses adapt to conducting business online, but travel to exotic holiday destinations might just be a matter of deferred demand awaiting release from shelter-in-place directives. 

Alternative Data: Creative exploration of existing alternative data sources can sometimes reveal unexpected proxy signals for individual and group behaviors. Combining and analyzing signals from multiple geolocation-enabled applications on a cell phone offers a somewhat crude picture of social distancing; combining cell-phone data with other sources, however, can give a more nuanced view that will also support a rigorous understanding of economic behaviors during a crisis. For example, predictive signals found in shipping traffic, purchase transactions, trucking and transportation, and satellite imaging data, can be combined to create new predictive models of consumer behaviors.

Health-Care Models: Based on known models of other infections such as influenza, quantitative analysts could create both a baseline for provider groups to understand how many COVID-19 patients they should be seeing and a model to predict how many they will be seeing as the pandemic progresses. Such models could help health-care and government institutions better understand where they are in the pandemic cycle—e.g., how many “weeks behind New York City or Italy”—in order to adjust behaviors, resources, and responses accordingly. 

If history is anything to go by, it’s not difficult to predict that what we are all enduring right now will result in significant and lasting changes to both society and consumer behavior. The Spanish Flu epidemic of 1918, coinciding with World War I, led to mass urbanization in the United States and to individuals giving up farm work and entering the manufacturing and service industries. Today, as the world reacts to slow the pandemic, advanced predictive analytics techniques (such as those originally developed for implementation in areas like quantitative finance) can reveal emerging patterns that may give us clues about what our post-COVID society will look like and point us to the areas that should be enhanced to better protect our communities in the future.