Source-linked AI summary
Characterizing Diabetes, Diet, Exercise, and Obesity Comments on Twitter
Amir Karami, Alicia A. Dahl, Gabrielle Turner-McGrievy, Hadi Kharrazi, Jr., George Shaw
TL;DR
Earlier Twitter studies examined topics within one health issue, while this study analyzes public discussions of diabetes, diet, exercise, and obesity. Its computational content-analysis framework identifies connections among these topics and the broader themes discussed alongside them.
Problem
Earlier Twitter studies extracted common topics for one health issue, leaving this study to characterize public opinions across diabetes, diet, exercise, and obesity at the population level.
Source: §1 · p. 2Method
The study uses a computational content-analysis framework to analyze Twitter discussions of four health issues, including lexicon-based LIWC analysis to identify health-related topics.
Source: §1 · p. 2, §2.3 · p. 4Results
Twitter discussions linked exercise with brain benefits, diabetes, meal planning, and obesity, while diabetes and obesity subtopics included other diseases, children, Alzheimer’s, and yoga.
Source: §4 · pp. 7–8Takeaways & Limitations
The framework is presented as applicable to other public-health issues, with possible use in physician-monitored communication devices and social-media weight-management interventions.
Source: §4 · pp. 7–8The analysis covered tweets from one month, although public opinion can change during a year.
Source: §4 · p. 8
Abstract
from arXiv · showhide
Social media provide a platform for users to express their opinions and share information. Understanding public health opinions on social media, such as Twitter, offers a unique approach to characterizing common health issues such as diabetes, diet, exercise, and obesity (DDEO), however, collecting and analyzing a large scale conversational public health data set is a challenging research task. The goal of this research is to analyze the characteristics of the general public's opinions in regard to diabetes, diet, exercise and obesity (DDEO) as expressed on Twitter. A multi-component semantic and linguistic framework was developed to collect Twitter data, discover topics of interest about DDEO, and analyze the topics. From the extracted 4.5 million tweets, 8% of tweets discussed diabetes, 23.7% diet, 16.6% exercise, and 51.7% obesity. The strongest correlation among the topics was determined between exercise and obesity. Other notable correlations were: diabetes and obesity, and diet and obesity DDEO terms were also identified as subtopics of each of the DDEO topics. The frequent subtopics discussed along with Diabetes, excluding the DDEO terms themselves, were blood pressure, heart attack, yoga, and Alzheimer. The non-DDEO subtopics for Diet included vegetarian, pregnancy, celebrities, weight loss, religious, and mental health, while subtopics for Exercise included computer games, brain, fitness, and daily plan. Non-DDEO subtopics for Obesity included Alzheimer, cancer, and children. With 2.67 billion social media users in 2016, publicly available data such as Twitter posts can be utilized to support clinical providers, public health experts, and social scientists in better understanding common public opinions in regard to diabetes, diet, exercise, and obesity.
1. Introduction
Despite reported links among diabetes, diet, exercise, and obesity, public perceptions have been examined mainly through surveys. This study uses Twitter to characterize public opinions and their relationships across all four health issues.
1. Introduction
- By 2014, more than 1.9 billion adults were overweight and over 600 million were obese, after global obesity prevalence doubled since 1980.Source: §1 · p. 1
- Studies report strong DDEO relations, while public-perception evidence remains survey-based.Source: §1 · p. 2
- Twitter posts offer a way to examine health opinions at population scale beyond surveys, interviews, and focus groups.Source: §1 · p. 2
- The study analyzes opinions across four health issues, extending prior Twitter topic studies that focused on one health issue at a time.Source: §1 · pp. 2–3
2. Methods
The framework combines collection, topic discovery, and topic-content analysis to characterize DDEO opinions in tweets.
2. Methods
- The three-phase framework combines data collection, topic discovery, and topic-content analysis using semantic and linguistic analyses.Source: §2 · p. 3
2.1. Data Collection
- Approximately 4.5 million English tweets were collected during June 2016 using pre-defined DDEO-related queries and Twitter’s real-time API stream.The paper states that its real-time collection sampled 10% of publicly available English tweets.Source: §2.1 · p. 3
2.2. Topic Discovery
- LDA identifies topics and themes in the tweets by grouping semantically related words and assigning words to topics with varying membership.Source: §2.2 · p. 4
- The analysis used Mallet LDA with default settings, removed stop words, and selected 425 topics using log-likelihood on 80% training and 20% test tweets.Source: §2.2 · p. 4
2.3. Topic Content Analysis
- LIWC’s health dictionary was used to identify which topics contained health-related words within a lexicon-based content analysis.Source: §2.3 · p. 4
3. Results
Among 4.5 million DDEO-related tweets, obesity was the largest category and diabetes the smallest. Topic analysis found overlapping DDEO subtopics and the strongest reported correlation between exercise and obesity.
3. Results
- 51.7% of tweets discussed obesity, compared with 8.0% diabetes, 23.7% diet, and 16.6% exercise.Source: §3 · p. 4
- Common subtopics ranged from blood pressure and heart attack for diabetes to vegetarian and mental-health themes for diet, brain and computer games for exercise, and cancer and children for obesity.DDEO terms also appeared as subtopics across the four categories.Source: §3 · p. 5, Table 2 · p. 6
- 80 of 222 health-related topics concerned exercise versus 21 diabetes; diet and obesity had 58 and 63, respectively.Source: §3 · p. 5, Table 2 · p. 6
- Exercise and obesity had the strongest reported correlation (p < .0002), followed by diabetes and obesity (p < .0005) and diet and obesity (p < .001).Source: Figure 2 · p. 6, §3 · p. 6
4. Discussion
The discussion interprets Twitter posts as a way to examine public opinions across DDEO topics and their subtopics, while noting important limits in the data. It connects these findings to potential uses in public health and clinical settings.
4. Discussion
- Twitter can capture varied user-expressed opinions, while traditional polls and surveys are limited by small sample sizes.Prior studies cited by the authors found strong correlation between Twitter health conversations and CDC statistics.Source: §4 · pp. 6–7
- The computational content-analysis framework examines large tweet datasets and surfaces health subtopics spanning diseases and personal experiences.The authors say the approach may be applied to other public health issues.Source: §4 · p. 7
- Diabetes subtopics linked it with diet, exercise, and obesity, as well as heart attack, Alzheimer, and yoga-related mental health benefits.The authors note that Alzheimer also appeared under obesity, prompting discussion of links among obesity, diabetes, and Alzheimer’s disease.Source: §4 · p. 7
- Diet had the most subtopics, including religious fasting and celebrity diet plans; other religions’ fasting was not captured in this collection.The authors suggest religious-diet findings may inform religious scholars considering health and psychological conditions during fasting.Source: §4 · pp. 7–8
- Exercise subtopics connected it with brain benefits, diabetes, meal planning, obesity, and computer games, including Pokémon Go.The authors suggest augmented-reality games may have potential to increase physical activity.Source: §4 · p. 8
- Obesity had the fewest subtopics; these included diabetes, children, and Alzheimer, which the authors connect to existing research on chronic-disease links.The authors also note public discussion of possible links between obesity and Alzheimer’s disease.Source: §4 · p. 8
- The authors identify possible uses in physician-monitored patient communication, social-media weight-management interventions, and population-wide health initiatives.The proposed applications address healthy behaviors and prevention related to diabetes, diet, exercise, and obesity.Source: §4 · p. 8
- The analysis omits users’ locations, uses a limited query set, and covers one month without tracking individuals over time.These limits constrain geographic comparisons and detection of changes in common themes over time.Source: §4 · p. 8
5. Conclusion
The study presents a transdisciplinary process for collecting, analyzing, and interpreting DDEO-related social-media posts. It frames this as an initial step toward more streamlined methods for identifying public-health trends in real time.
5. Conclusion
- The study presents a first step toward routine collection, analysis, and interpretation of DDEO-related social-media posts using a transdisciplinary approach.The authors state that more streamlined and robust methods could allow faster identification of public-health trends in real time.Source: §5 · p. 9