Logistic regression analysis of textual data on suicidal ideation

Suicide prevention requires careful consideration of the entire process, including ideation, attempt, and death. Understanding factors related to suicidal ideation is particularly critical for early intervention and effective prevention measures. In this study, we identified such factors by applying logistic regression analysis to textual data posted in September 2024 on NHK’s “Facing Suicide” website, where individuals shared their thoughts and feelings. Several keywords, including “self/identity,” were grouped into five categories, while demographic attributes such as gender, age, day of the week, and time of day were also examined. This analysis found that demographic and temporal factors influenced message content. Women and younger individuals were more likely to post messages centred on “self/identity” and “functional words/actions,” suggesting that concerns related to self-perception and existence became more pronounced during late-night hours. This temporal pattern indicates that nighttime may serve as a critical period for heightened suicidal ideation.Author summary: The author specializes in computational and spatial statistics, focusing on the visualization and analysis of epidemiological data—particularly related to suicide. They apply advanced statistical methods and open-source software, such as R, to construct disease maps that reveal spatial patterns in suicide mortality. This approach enables the identification of high-risk hotspots and safer coolspots, providing valuable insights into local factors influencing mental health outcomes. Beyond mapping, the author has contributed as an epidemiological expert to global health initiatives, including World Health Organization suicide reports, thereby informing prevention strategies at an international level. Their research also examines the relationship between suicide risk and temporal and lifestyle factors, comparing trends in Japan before and after the COVID-19 pandemic. By incorporating generative AI into evidence-based policy-making (EBPM), they strive to refine analytical frameworks and support informed decision-making. The author aims to drive effective interventions in future mental health policy initiatives through this blend of computational tools, spatial analysis, and public health expertise.