Good Analytics INC — Department of Health and Human Services SBIR Phase I: ODCDC

Description

Population public health surveillance for scientific inquiry evidenced based regulatory decision making planning objects such as Healthy People and allocating clinical care remains largely unchanged over the past years Monitoring social media to fill surveillance gaps is promising but a major barrier is selection bias tweeters do not represent the US population Public health broadly has yet to address this bias This proposal develops Health Cohort a cohort purposively sampled to match the US population demographic profile Just as demographic probability sampling serves as basis of validity for surveys this purp osive selection of Twitter accounts should yield valid national and demographically specific insights This proposal will identify public Twitter accounts removing bot and other spurious accounts infer the location and other demographics associated with each account using custom computer software select a large cohort that reflects US Census distributions To demonstrate the utility of this methodological solution we will analyze tweets about tobacco the example used in the funding solicitation In particular these results can inform the CDC s surveillance such as questions for the Behavioral Risk Factor SuNeillance System the content of CDC lead anti smoking campaigns such as the groundbreaking Tips from Former Smokers Campaign and potentially clinically relevant outcomes