Ai+ Training

Visualization and epidemiology have always advanced each other symbiotically, like a unique cross-disiplinary lichen, for as long as both have existed. Ever since John Snow's 1854 Broad Street cholera map ushered in the modern practice of both fields, it has served as an important cornerstone in guiding their advancement. The expectations for epidemiological visualizations however have grown in the centuries since that foundational graphic as much as each requisite field has grown complex. And perhaps most pernicious, the sophistication and gravity of disease outbreaks has continuously evolved as modern society has become inextricably interconnected.

The world has been experiencing the devastating result of this sophistication first hand as it has been battling the COVID-19 pandemic for the better part of a year now. What used to be one inferential tool in an expert's arsenal to investigate the spread of a disease has now become a public health staple. Essential for promoting prosocial collective action (i.e. social distancing, wearing a mask, etc.) necessary for mitigating a pandemic, data visualization has seen its breakthrough moment as it has played an outsized role in communicating the severity and impact of COVID-19.

Using existing public COVID-19 visualizations and predictive models as case studies, I will ground effective visual design in cognitive psychology and discuss the nuance of communicating complex model results to a lay audience. This talk will focus specifically on the effectiveness of various techniques for showing change over time, visualizing uncertainty in data, and communicating model predictions. Code examples using Vega-Lite will be presented but no previous data visualization (or epidemiological) knowledge is assumed or necessary to get value out of this talk.