Population attitudes toward contraceptive methods over time on a social media platform.

摘要: Background Contraceptive method choice is often strongly influenced by the experiences and opinions of one’s social network. Although media, including Twitter, increasingly influences reproductive-age individuals, discussion contraception in this setting has yet to be characterized. Natural language processing, a type machine learning which computers analyze natural data, enables analysis. Objective This study aimed illuminate temporal trends attitudes toward long- short-acting reversible contraceptive methods tweets between 2006 2019 establish media platforms as alternate data sources for large-scale sentiment analysis on contraception. Study Design We studied English-language mentioning prescription March (founding Twitter) December 2019. Tweets were extracted using search terms, generic or brand names, colloquial abbreviations. characterized performed tweets. used Mann-Kendall nonparametric tests assess overall number positive, negative, neutral referring each method. The code reproduce available at https://github.com/hms-dbmi/contraceptionOnTwitter . Results 838,739 least 1 annual contraception-related increased considerably over period. intrauterine device was most commonly referenced (45.9%). Long-acting mentioned more than ones (58% vs 42%), proportion long-acting time. In single (n=665,064), greatest all negative (65,339 160,713 with 95% confident sentiment, 40.66%). nearly twice likely positive compared (19.65% 10.21%; P Conclusion Recognizing influence networks decision making, may useful collection dissemination information about