Tweet topics on cancer among Indian Twitter users—computational approach using latent Dirichlet allocation topic modelling

Understanding the extent and content of conversations on cancers inform the stakeholders regarding the needs of the community in terms of knowledge, support and interventions. This study identified the topics of tweet content shared regarding cancer, source of messages and the degree of reachability of identified topics among Twitter users in India. Twitter messages geocoded within India, related to cancer and posted between September 15, 2021 and October 15, 2021 were retrieved using the Twitter application programming interface based on keywords identified from Symplur Signals. The tweets were pre-processed to remove the stop words, hashtags and Uniform Resource Locators. Tweets were visualized using word clouds and correlations between word tokens. Latent Dirichlet allocation (LDA) topic model, an unsupervised machine learning technique was used to identify the commonly discussed cancer topics. A total of 6374 tweets from 3135 unique twitter users were analysed in the study. Majority of the tweets (60.8%) were from the individual twitter users. LDA model identified four topics: (1) prevention, early detection and promotion (36.1%), (2) seeking support and sharing personal experience (15.8%), (3) Human Papillomavirus vaccine and cancer research (13.4%), (4) risk factors, treatment and raising awareness (34.7%). Among the four identified topics, prevention, early detection and promotion had the highest reachability. Twitter is being used as a potential alternative communication platform for disseminating cancer-related information in India. The topics identified in the study provides useful insights for public health professionals and organizations for aligning cancer-related engagement and education for the target audience.