Traditional health research often faces limitations in scale, scope, timeliness and costs. Data is usually collected through medical exams, questionnaires, surveys and laboratory tests which often are time consuming and costly. It’s especially laborious for patients, who have to take up their time, travel and add an extra burden by participating in such studies.
Digital health research is the use of modern technologies to improve healthcare outcomes. It lies at the exciting intersection of medical research, computer science, and data analytics, offering ways to better understand, diagnose, and treat diseases. It harnesses the wealth of data generated through digital platforms (e.g. social media activity, wearable devices) that were not specifically generated for health research purposes also called the digitosome. Researchers can thus gain insights into lifestyle patterns, disease progression, and develop pharmacovigilance strategies. In this project, we focus on using social media.
Social media refers to digital platforms that allow users to create, share, and interact with each other. These platforms serve as virtual communities where individuals can connect with others, either publicly or within specific networks. Particularly, social media are used by health researchers in different ways:
Twitter is a social media company founded in 2006 that is now called X since mid-2023. It is designed for microblogging, allowing users to post and interact with messages known as “tweets” which are limited to 280 characters for classic users and longer for premium users.
On Twitter (or X), engaged online communities of patients have emerged. Supportive networks for individuals living with diabetes were created with, for instance the GBDoc (Great britain Diabetes online community) and DSMA (Diabetes Social Media Advocacy) groups. These communities are valuable for sharing experiences, seeking guidance and fostering connections among people facing similar challenges related to diabetes management. Hashtags such as #GBDoc, #DSMA, #DiabetesCommunity, and #T1D (Type 1 Diabetes) serve as digital markers, enabling users to find and participate in conversations on topics like blood glucose monitoring, insulin dosing, dietary considerations, exercise routines, emotional well-being, and the latest advancements in diabetes research and treatment. Such communities generate tweets that are of interest to health research, as they provide direct access to what’s happening in people with diabetes everyday life.
A previous study called “Insulin pricing and other major diabetes-related concerns in the USA: a study of 46,407 tweets between 2017 and 2019”, also part of the WDDS project, serves as a basis for our research. This study used Twitter data to better understand the concerns and emotional issues of people with diabetes in the US. Machine learning techniques were used to filter tweets published by people with diabetes (removing institutional content and jokes), geolocate them in the USA and identify emotional elements. The findings revealed 30 topics of interest among these tweets. A significant proportion of these were focusing on insulin pricing with both positive emotions (joy, love) referring to advocacy for affordable insulin and sadness related to the frustration of insulin prices. 5 clusters were related to solidarity and support with a majority of positive emotions. The most negative topics were related to diabetes distress (with anger, sadness and fear), to diabetic and insulin shock (anger, fear) and comorbidities (sadness).
Based on this work, we aim to extend the analysis globally to understand the complexities of diabetes burden around the world. While the initial study was confined to the United States, the WDDS project seeks to create a global picture of diabetes distress and its multifaceted impact on people’s lives. We are also developing robust methodologies to analyze Twitter data for health research, thereby enabling a more nuanced understanding of public health issues.
