Analyzing Online Media Articles on Diabetes using Natural Language Processing: A Comparative Study of Indian Ocean Region and France

Abstract

Background: Diabetes is a global health concern affecting millions of people worldwide. However, knowledge, attitudes, and practices related to this disease vary widely across different regions. This article aims to investigate mediainfluenced perceptions about diabetes in France and the Indian Ocean countries using natural language processing (NLP) techniques applied to online news articles. Findings aims to provide expert in Health Literacy (HL) and health promotion to develop better communication strategies. Method: Constitute a datatset of Online news articles on Diabetes and apply NLP like Word2Vec for word integration, LDA for topic identification, and transformer-based classification models (e.g., BERT and its variants) for sentiment analysis. processing (NLP). Results: Sentiment analysis revealed more negative discussions about diabetes in the Indian Ocean region (48%) compared to France (32%), with neutral articles dominating in France (42%). In terms of topic Identification there were some topic which appeared for France which were not present for indian ocean region. Discussions: The findings of this study indicate that perceptions and discussions about diabetes differ between two regions, which have implications for public health interventions and communication strategies. However, the study is limited by the initial amount of information captured for analysis