BACKGROUND
Physical activity is implicated in preventing many chronic diseases. However, studies show, that the recommendations of the WHO regarding physical activity are rarely met. Many different theories to promote physical activity are applied globally. Most of these theories are based on expert opinion or results from clinical or experiment settings. The application of said theories has limited success however. This raises the question, whether these theories actually reflect what the general population deems important.
OBJECTIVE
To investigate popular online topics or themes related to physical activity by analyzing search queries and social media discussions. Furthermore, we provide a large dataset of 4 million tweets and an extensive machine learning toolbox for exploratory research adaptable to any research question.
METHODS
We collected data on the most searched topics regarding "physical activity" in Google searches worldwide over the past 15 years, and conversations in the complete Twitter repository from April 1st, 2006 (Twitter's launch) to April 1st, 2023. To analyze the Twitter data, we utilized deep learning natural language processing techniques. Searches and analyses were performed in 2023 at the University of Bern, Switzerland.
RESULTS
The most frequently searched topics on Google related to "physical activity" include its definition, as well as health, fitness, and benefits. Among the most mentioned positive terms are "health," "sport," and "exercise," which were mentioned twice as often as negative context mentions. Twitter data sorted by sentiment shows clear patterns emerging in word frequencies regarding motivation for, and problems connected to, physical activity.
CONCLUSIONS
People across the English-speaking world are primarily interested in understanding what "physical activity" entails. Therefore, we recommend that policymakers and professionals in the field prioritize providing clear and straightforward definitions and guidelines for the general population. Additionally, we encourage the use of real-world data to inform the modeling of physical activity patterns and behaviors. This approach can better align with the interests and needs of the general population, as reflected in their online search and discussion behaviors.