Publication

Abstract : It is estimated that 1 out of 77 people in India are formally diagnosed with diabetes, which is alarming. Nowadays, people use social media platforms such as Facebook and Twitter to express their beliefs, feelings, and concerns on any topic, including healthcare. It is interesting and highly useful to analyze this humongous data to unearth latent patterns for further analysis and decision-making. This study aimed at (1) analyzing the sentiment of the diabetes-related tweets from India and (2) unearthing latent themes of discussion about diabetes in India from Twitter. We collected tweets from India using the keywords - diabetes, diabetes mellitus, and hyperglycemia. We identified the sentiments of retrieved tweets using a deep learning sentiment classifier followed by a topic modeling approach to analyze the major themes of discussion. The results show that the majority of the sentiments were positive, but still, there are concerns regarding diabetes risk and the alarming rate of diabetes in India. Our topic modeling results showed some interesting patterns from the discussions, such as diabetes advisory, diabetes risk, diabetes warning, and diabetes control.