TY  - JOUR
AU  - Metwally, Omar
AU  - Blumberg, Seth
AU  - Ladabaum, Uri
AU  - Sinha, Sidhartha R
PY  - 2017
DA  - 2017/06/07
TI  - Using Social Media to Characterize Public Sentiment Toward Medical Interventions Commonly Used for Cancer Screening: An Observational Study
JO  - J Med Internet Res
SP  - e200
VL  - 19
IS  - 6
KW  - Twitter
KW  - sentiment analysis
KW  - cancer screening
KW  - colonoscopy
KW  - mammography
KW  - Pap smear
KW  - Papanicolaou test
KW  - social media
KW  - early detection of cancer
AB  - Background: Although cancer screening reduces morbidity and mortality, millions of people worldwide remain unscreened. Social media provide a unique platform to understand public sentiment toward tools that are commonly used for cancer screening. Objective: The objective of our study was to examine public sentiment toward colonoscopy, mammography, and Pap smear and how this sentiment spreads by analyzing discourse on Twitter. Methods: In this observational study, we classified 32,847 tweets (online postings on Twitter) related to colonoscopy, mammography, or Pap smears using a naive Bayes algorithm as containing positive, negative, or neutral sentiment. Additionally, we characterized the spread of sentiment on Twitter using an established model to study contagion. Results: Colonoscopy-related tweets were more likely to express negative than positive sentiment (negative to positive ratio 1.65, 95% CI 1.51-1.80, P<.001), in contrast to the more positive sentiment expressed regarding mammography (negative to positive ratio 0.43, 95% CI 0.39-0.47, P<.001). The proportions of negative versus positive tweets about Pap smear were not significantly different (negative to positive ratio 0.95, 95% CI 0.87-1.04, P=.18). Positive and negative tweets tended to share lexical features across screening modalities. Positive tweets expressed resonance with the benefits of early detection. Fear and pain were the principal lexical features seen in negative tweets. Negative sentiment for colonoscopy and mammography spread more than positive sentiment; no correlation with sentiment and spread was seen for Pap smear. Conclusions: Analysis of social media data provides a unique, quantitative framework to better understand the public’s perception of medical interventions that are commonly used for cancer screening. Given the growing use of social media, public health interventions to improve cancer screening should use the health perceptions of the population as expressed in social network postings about tests that are frequently used for cancer screening, as well as other people they may influence with such postings. 
SN  - 1438-8871
UR  - http://www.jmir.org/2017/6/e200/
UR  - https://doi.org/10.2196/jmir.7485
UR  - http://www.ncbi.nlm.nih.gov/pubmed/28592395
DO  - 10.2196/jmir.7485
ID  - info:doi/10.2196/jmir.7485
ER  -