Using Google Cloud and Predictive Analytics to Promote Influenza Vaccination among People with Diabetes

Background: Although influenza vaccination is recommended for people with type 1 and 2 diabetes, uptake is suboptimal, often leading to complications and hospitalizations. Vaccine acceptance is influenced by several psychosocial and behavioral factors, including one’s knowledge of flu and vaccination beliefs. We used novel data sources to understand predictors of vaccination adherence among countries and subpopulations.

Methods: We used VaxiTrends, a proprietary dataset, to identify variables influencing vaccine uptake among adults in the U.S., China, France, Mexico, and U.K. We combined the data with NHI Interview Survey results, which offer insights into health behaviors of Americans with diabetes. This information was aggregated with data from Google’s FluTrend, which signals flu incidence, and Google Trends Application Programming Interface, which tracks search keyword popularity by location and date in a de-identified fashion. Cluster analyses indicated factors most positively and negatively associated with acceptance and adherence.

Results: Our analysis indicates that targeted interventions focussing on country-specific drivers of flu vaccination adherence among subpopulations of people with diabetes could lead to an increase in vaccination rates of up to 5%. In France, Mexico and U.K., relationships with HCPs were most predictive of vaccination adherence, while in the U.S., reminders and strategies addressing affordability concerns were key predictors. In China, encouragement from relatives, convenient options for receiving vaccinations and education to counter fear of flu vaccination may have the greatest impact. Within countries,we found some patient subgroups were more amenable to vaccination than others.

Conclusion: By identifying country-specific behavioral drivers of flu vaccination and most receptive subgroups, we can design more tailored interventions to increase uptake.