Applying Sparse Machine Learning Methods to Twitter: Analysis of the 2012 Change in Pap Smear Guidelines. A Sequential Mixed-Methods Study

Abstract

BackgroundIt is difficult to synthesize the vast amount of textual data available from social media websites. Capturing real-world discussions via social media could provide insights into individuals’ opinions and the decision-making process.ObjectiveWe conducted a sequential mixed methods study to determine the utility of sparse machine learning techniques in summarizing Twitter dialogues. We chose a narrowly defined topic for this approach: cervical cancer discussions over a 6-month time period surrounding a change in Pap smear screening guidelines.MethodsWe applied statistical methodologies, known as sparse machine learning algorithms, to summarize Twitter messages about cervical cancer before and after the 2012 change in Pap smear screening guidelines by the US Preventive Services Task Force (USPSTF). All messages containing the search terms “cervical cancer,” “Pap smear,” and “Pap test” were analyzed during: (1) January 1–March 13, 2012, and (2) March 14–June 30, 2012. Topic modeling was used to discern the most common topics from each time period, and determine the singular value criterion for each topic. The results were then qualitatively coded from top 10 relevant topics to determine the efficiency of clustering method in grouping distinct ideas, and how the discussion differed before vs. after the change in guidelines .ResultsThis machine learning method was effective in grouping the relevant discussion topics about cervical cancer during the respective time periods (~20% overall irrelevant content in both time periods). Qualitative analysis determined that a significant portion of the top discussion topics in the second time period directly reflected the USPSTF guideline change (eg, “New Screening Guidelines for Cervical Cancer”), and many topics in both time periods were addressing basic screening promotion and education (eg, “It is Cervical Cancer Awareness Month! Click the link to see where you can receive a free or low cost Pap test.”)ConclusionsIt was demonstrated that machine learning tools can be useful in cervical cancer prevention and screening discussions on Twitter. This method allowed us to prove that there is publicly available significant information about cervical cancer screening on social media sites. Moreover, we observed a direct impact of the guideline change within the Twitter messages.

Representations of Codeine Misuse on Instagram: Content Analysis

“…At a stage where large-scale image analysis is still evolving, we feel that even this initial content analysis is informative for the public health community. Our study is in line with the multiphase approach used in other similar Internet-based research wherein preliminary qualitative analysis on small samples [ 38 , 39 ] are used to inform large-scale automated methodologies [ 40 , 41 ] such as machine learning. However, despite the prevalence of misuse, the paucity of data on this topic requires that any intermediate findings be tested further before deemed useful for the development of much-needed interventions to prevent uptake and curb misuse.…”

Lung Cancer Messages on Twitter: Content Analysis and Evaluation

“…Early detection Text that describes screening tests (eg, LDCT) and family history [4]. needed in light of recent announcements about updated screening recommendations, novel treatment approaches, scientific discoveries, and celebrity cancer diagnoses [3,[6][7][8][9][10]. Preliminary research suggests that social media may serve as an important tool for disseminating prevention, screening, treatment, and survivorship messages to a widespread audience [3,7,9,11].…”

“…However, the nature and extent of such cancer-related communication at different points in the cancer control continuum is unknown. Recent research has examined the use of social media, in particular Twitter, to identify how patients communicate about their screening experiences (eg, for mammography) [12], respond to changing US Preventive Services Task Force guidelines [1,8], assess the source and credibility of colorectal cancer information [13], how providers recruit for clinical trials [14], and how cancer survivors tweet about their journeys [7,15]. However, there is a paucity of research evaluating how site-specific cancer communication may attend to audience needs for information ranging from prevention through end-of-life considerations.…”

“…Messages were manually coded for their place in the cancer continuum (eg, diagnosis, survivorship) [4,5] and other key types content (eg, cues to action, research study) that were identified from previous research on cancer messages on Twitter [8,18] and our previous research examining Twitter messages in disaster scenarios [19] and emerging infectious disease contexts [20], and an initial, exploratory coding of a subset of 150 messages. After a brief training session, three coders then coded 100 randomly selected lung cancer-related messages from the data set to establish intercoder reliability (ICR).…”

Using Social Media to Target Cancer Prevention in Young Adults: Viewpoint

“…Myriad examples of this type of work exist across disparate public health domains including substance use [ 15 ], body weight-associated stigma [ 16 ], and infectious disease surveillance [ 11 , 17 ]. For example, Lyles et al performed this observational type of analysis for cervical cancer prevention discussions among young women on Twitter [ 18 ]. The analysis demonstrated that women do share publicly their experiences with cervical cancer screening, often with language encouraging peers to undergo screening as well.…”