We analyzed 34 million tweets from 172 countries to get a better understanding of what people with diabetes are experiencing worldwide.
The majority of tweets were from North America, followed by Europe and Central Asia. We collected fewer tweets from regions like the Middle East and North Africa. Figure 1 shows the repartition of the tweets included for analysis.
The topics of interest varied significantly across regions. In North America, for instance, a significant number of tweets were about the high cost of insulin, reflecting public sentiment about healthcare systems. However, some topics had universal resonance: ‘Glycemic Control,’ ‘Family and Relatives,’ and ‘Food Choices’ were discussed across six out of seven regions analyzed. Sentiment analysis revealed an overall emotion of sadness, especially when discussing topics like daily diabetes management and medication costs. However, tweets that offered advice, motivational messages, or peer support had more positive sentiments. In fact, South Asia had the highest proportion of positive tweets, while Latin America and the Caribbean had the most negative ones.
In all, diabetes is a global issue that requires nuanced, region-specific attention. The tweets we analyzed indicated that while there are universal concerns, there are also unique regional challenges that healthcare systems must address. By focusing on these, we can move to a more patient-centric approach in diabetes care, providing tailored solutions that consider both the emotional and physical well-being of people with diabetes.
Reference: Global diabetes burden: analysis of regional differences to improve diabetes care
Cohort studies are a research design where a group of people, known as a cohort, are studied over a period of time to observe how certain exposures or characteristics are linked to specific outcomes. Thus, just as traditional cohort track people’s health over time to understand diseases better, virtual digital cohort studies track social media to capture what people are saying about their health conditions through time. It gives us quick access to large numbers of real-life experiences, allowing us to identify trends and issues we might not notice otherwise. For instance, by studying tweets from people talking about their own diabetes, we can spot recurring challenges that happen during their daily life. And unlike traditional studies, where participants might give answers they think their doctors want to hear, social media offers raw, unfiltered insights. This method is not perfect and cannot replace traditional medical studies, but it adds a new layer to our understanding by focusing on what patients are actually experiencing and talking about.
In a healthcare world where reducing patient burden and reducing costs are important challenged, we’ve created a solution that addresses both regarding cohort studies. We introduced ALTRUIST, an open-source Python package designed for this specific research paradigm. The acronym ALTRUIST stands for “virtuAL digiTal cohoRt stUdy on TwItter uSing pyThon,” mimicking traditional cohort studies using digital data. It reduces both the time and financial resources needed to generate valuable insights, proving to be an excellent complement to traditional epidemiological methods. ALTRUIST was developed in Python 3.8. It not only offers an alternative to traditional epidemiological research but also augments the depth of studies by tapping into real-world conversations among patients on platforms like Twitter. ALTRUIST was developed following the different stages of traditional cohort studies, from data collection to preprocessing and analysis, in an automated flow. This also involves leveraging AI models for text analysis and survival analysis. Figure 2 shows the structure and workflow of the package.
Find more information: ALTRUIST
The Worldwide Online Digital Health Observatory serves as a comprehensive platform for real-time health-related data analysis. By leveraging social media inputs, specifically Twitter, the observatory captures a broad spectrum of global health conversations. Interactive maps and word clouds offer visual representations of the geographic distribution and frequency of health-related topics. Additional features such as cluster analysis categorize tweets into related topics are completed with sentiment analysis scores. This platform aims to be a resource for public health agencies, researchers, and policy makers, offering actionable insights into global health trends and concerns. You can play with this observatory here.
