Association Between HIV-Related Tweets and HIV Incidence in the United States: Infodemiology Study.
HIV/AIDS
natural language processing
social media
surveillance
youth
Journal
Journal of medical Internet research
ISSN: 1438-8871
Titre abrégé: J Med Internet Res
Pays: Canada
ID NLM: 100959882
Informations de publication
Date de publication:
24 06 2020
24 06 2020
Historique:
received:
25
11
2019
accepted:
22
03
2020
revised:
05
03
2020
entrez:
25
6
2020
pubmed:
25
6
2020
medline:
17
12
2020
Statut:
epublish
Résumé
Adolescents and young adults in the age range of 13-24 years are at the highest risk of developing HIV infections. As social media platforms are extremely popular among youths, researchers can utilize these platforms to curb the HIV epidemic by investigating the associations between the discourses on HIV infections and the epidemiological data of HIV infections. The goal of this study was to examine how Twitter activity among young men is related to the incidence of HIV infection in the population. We used integrated human-computer techniques to characterize the HIV-related tweets by male adolescents and young male adults (age range: 13-24 years). We identified tweets related to HIV risk and prevention by using natural language processing (NLP). Our NLP algorithm identified 89.1% (2243/2517) relevant tweets, which were manually coded by expert coders. We coded 1577 HIV-prevention tweets and 17.5% (940/5372) of general sex-related tweets (including emojis, gifs, and images), and we achieved reliability with intraclass correlation at 0.80 or higher on key constructs. Bivariate and multivariate analyses were performed to identify the spatial patterns in posting HIV-related tweets as well as the relationships between the tweets and local HIV infection rates. We analyzed 2517 tweets that were identified as relevant to HIV risk and prevention tags; these tweets were geolocated in 109 counties throughout the United States. After adjusting for region, HIV prevalence, and social disadvantage index, our findings indicated that every 100-tweet increase in HIV-specific tweets per capita from noninstitutional accounts was associated with a multiplicative effect of 0.97 (95% CI [0.94-1.00]; P=.04) on the incidence of HIV infections in the following year in a given county. Twitter may serve as a proxy of public behavior related to HIV infections, and the association between the number of HIV-related tweets and HIV infection rates further supports the use of social media for HIV disease prevention.
Sections du résumé
BACKGROUND
Adolescents and young adults in the age range of 13-24 years are at the highest risk of developing HIV infections. As social media platforms are extremely popular among youths, researchers can utilize these platforms to curb the HIV epidemic by investigating the associations between the discourses on HIV infections and the epidemiological data of HIV infections.
OBJECTIVE
The goal of this study was to examine how Twitter activity among young men is related to the incidence of HIV infection in the population.
METHODS
We used integrated human-computer techniques to characterize the HIV-related tweets by male adolescents and young male adults (age range: 13-24 years). We identified tweets related to HIV risk and prevention by using natural language processing (NLP). Our NLP algorithm identified 89.1% (2243/2517) relevant tweets, which were manually coded by expert coders. We coded 1577 HIV-prevention tweets and 17.5% (940/5372) of general sex-related tweets (including emojis, gifs, and images), and we achieved reliability with intraclass correlation at 0.80 or higher on key constructs. Bivariate and multivariate analyses were performed to identify the spatial patterns in posting HIV-related tweets as well as the relationships between the tweets and local HIV infection rates.
RESULTS
We analyzed 2517 tweets that were identified as relevant to HIV risk and prevention tags; these tweets were geolocated in 109 counties throughout the United States. After adjusting for region, HIV prevalence, and social disadvantage index, our findings indicated that every 100-tweet increase in HIV-specific tweets per capita from noninstitutional accounts was associated with a multiplicative effect of 0.97 (95% CI [0.94-1.00]; P=.04) on the incidence of HIV infections in the following year in a given county.
CONCLUSIONS
Twitter may serve as a proxy of public behavior related to HIV infections, and the association between the number of HIV-related tweets and HIV infection rates further supports the use of social media for HIV disease prevention.
Identifiants
pubmed: 32579119
pii: v22i6e17196
doi: 10.2196/17196
pmc: PMC7380998
doi:
Types de publication
Journal Article
Research Support, N.I.H., Extramural
Research Support, Non-U.S. Gov't
Langues
eng
Sous-ensembles de citation
IM
Pagination
e17196Subventions
Organisme : NIMHD NIH HHS
ID : T37 MD014248
Pays : United States
Informations de copyright
©Robin Stevens, Stephen Bonett, Jacqueline Bannon, Deepti Chittamuru, Barry Slaff, Safa K Browne, Sarah Huang, José A Bauermeister. Originally published in the Journal of Medical Internet Research (http://www.jmir.org), 24.06.2020.
Références
PLoS One. 2014 Apr 17;9(4):e91711
pubmed: 24743728
J Med Internet Res. 2014 Oct 06;16(10):e228
pubmed: 25289463
Psychol Sci. 2011 Feb;22(2):235-42
pubmed: 21177515
Am J Public Health. 2010 Apr 1;100 Suppl 1:S40-6
pubmed: 20147663
AIDS Behav. 2016 Jun;20(6):1256-64
pubmed: 26650382
JAMA. 2018 Dec 18;320(23):2417-2418
pubmed: 30428002
PLoS One. 2015 Oct 05;10(10):e0139701
pubmed: 26437454
J Sex Res. 2017 May -Jun;54(4-5):446-464
pubmed: 28135857
Science. 2014 Mar 14;343(6176):1203-5
pubmed: 24626916
Health Psychol. 2015 Dec;34S:1252-60
pubmed: 26651466
PLoS Med. 2008 Apr 8;5(4):e80
pubmed: 18590346
J Adolesc Health. 2015 Feb;56(2):139-45
pubmed: 25620299
J Stud Alcohol Drugs. 2016 Mar;77(2):349-53
pubmed: 26997194
Addict Behav. 2017 Feb;65:289-295
pubmed: 27568339
Am J Public Health. 2018 Oct;108(10):1378-1384
pubmed: 30138075
Prev Med. 2014 Jun;63:112-5
pubmed: 24513169
Am J Public Health. 2017 Jan;107(1):e1-e8
pubmed: 27854532