Author ORCID Identifier
Date of Award
12-2-2024
Document Type
Thesis (Ph.D.)
Department or Program
Quantitative Biomedical Sciences
First Advisor
Nicholas C. Jacobson
Abstract
More than 700,000 individuals across the globe die by suicide each year, making suicide a leading cause of death. To improve suicide risk assessment and prediction, addressing the limitations surrounding the conceptualization of suicidal thought and behavior (STB) is essential. Research within the past decade has laid the groundwork for the application of more complex models (e.g., machine learning and deep learning architectures) to probe STB phenomenology. Such work has tended to analyze content from ubiquitous social media and online communication platforms which, despite advantages related to data volume and density, nevertheless offer censored and incomplete representations of STB. As a result, the markers and models that stem from these efforts ultimately paint a constrained and artificial picture of STB expression. The current body of work sought to address this limitation by leveraging behavioral data that is derived from both social and personal contexts, emphasizing situations where STB can be expressed in manners that are ecologically valid, minimally censored, and temporally rich. To this end, this research first explored the public face of suicide by modeling the semantic and social dynamic properties of STB on the pro-choice suicide forum, Sanctioned Suicide. The written and temporal properties of private-facing suicidal expression were then interrogated through ecological momentary assessment-based diary entries of individuals with major depressive disorder as well as through online search engine queries of anonymous Internet users. Via the primary application of natural language processing, network analysis, and machine learning-based techniques, this research provides novel insights regarding the presentation of suicide and highlights potentially useful markers of STB expression that may serve to bolster progress toward more sophisticated prevention, intervention, and mitigation of STB within the modern digital arena.
Recommended Citation
Lekkas, Damien, "Misery's Company: Modeling the Public and Private Faces of Suicidal Thought and Behavior in the Digital Era" (2024). Dartmouth College Ph.D Dissertations. 328.
https://digitalcommons.dartmouth.edu/dissertations/328
Included in
Applied Statistics Commons, Data Science Commons, Quantitative Psychology Commons, Statistical Models Commons
