Author(s): Liam Ben Sadik, Jalynn Reeves
Mentor(s): Kevin Johnston
Institution UTech
The recent COVID-19 epidemic has illustrated the need for near real-time data reporting and analysis in incidence tracking, which is critical for allocation of resources and public information campaigns as infection rates vary. Beyond the disease itself, COVID-19 also created massive delays in supply chains, resulting in medication shortfalls, while the induced isolation (among other factors) resulted in increased usage of prescribed psychiatric medication. Recent studies have shown a strong correlation between google search trends, and public health variables such as COVID-19 incidence. In this project, we analyze the usefulness of google trends in predicting alterations in prescription drug demand. We utilize APIs to extract prescription drug search terms from google trends and compare those results with publicly available prescription usage from the MEPS database. We have already identified a considerable number (>50) of medications for which drug usage and google trend searches for the same prescription drug were highly correlated over time. Here we will present our results, including a more complete list of medications that show strong correlation with real word drug usage obtained from MEPS database, an analysis identifying common features of drugs whose usage correlates with google trends data, and our statistical and machine learning models trained to identify rapid alterations and anomalies in prescription drug usage, from the google trends dataset. Additionally, we will present an early version of our online web portal, designed to allow users, including public health officials and pharmaceutical analysis employees, to visualize, analyze and adjust strategies based on insight obtained from our models.