JMIR Publications

ABSTRACT

Background:

HIV disproportionately affects young men ages 13-24 in the United States. As social media platforms are extremely popular among youth and young adults, investigators have an opportunity to utilize these platforms to curb the HIV epidemic by examining discourse regarding HIV and its associations with epidemiological data for HIV prevention.

Objective:

The goal of this study is to examine how Twitter activity among young men is related to HIV incidence in the population.

Methods:

We characterized the HIV-related messages on Twitter from male social media users, aged 13-24 years, using integrated human-computer techniques. Using natural language processing (NPL), we identified tweets related to HIV risk and prevention. Our NLP algorithm identified 89.1% of relevant tweets, which were manually coded by expert annotators. We coded 1577 prevention tweets and 17% of general sex tweets (including emojis, gifs and images, n=940), achieving 80% inter-coder reliability on key constructs. Bivariate and multivariate analyses were used to identify spatial patterns in posting as well as relationships between posting and local HIV rates.

Results:

We analyzed 2517 tweets that were identified as relevant to HIV risk and prevention tags geolocated in 109 counties throughout the United States. After adjusting for region, HIV prevalence, and SDI, our findings indicate that HIV-specific tweets from non-institutional accounts were negatively associated with HIV incidence at the county level the following year (OR: 0.97, p=.038). After adjusting for covariates, every one hundred tweet increase HIV-specific tweets per capita from non-institutional accounts was associated with a 3% decrease in the incidence rate of HIV the following year in a given county.

Conclusions:

Twitter may serve as a proxy of public behavior related to HIV and the association with HIV infection rates further supports the use of social media for HIV prevention.