Suicide risk factors: A language analysis approach in social media

Suicide represents a public health issue that requires new preventive strategies. Therefore, this study analyzes differences in language use between a themed posts group (suicide and depression) and a random posts group (non-specific topics) from different social media platforms. In addition, the similarity of the texts of themed posts group with the set of phrases linked to suicide risk factors is analyzed. Texts were processed using the Linguistic Inquiry and Word Count (LIWC) software. A 95% bootstrap confidence interval (CI) was built for the difference in means for the resulting values per word category across groups. Significant differences in the use of language were observed between the themed post group and the random post group, whereas no differences were found between the themed post group and the set of phrases linked to suicide risk factors. These results can contribute to the development of risk-detection tools, which identify users at risk based on the analysis of language use embedded in social media. Such opportune detection would significantly increase the suicide prevention work. However, there is still a latent need for doing more research in this field, especially for the Mexican culture.