Aerospace engineer-turned-data scientist specializing in machine learning with a passion for developing data-driven solutions to societal issues and public health research.


PROJECTS


Jan 2024 – May 2024
Literature Review: On the Relationship Between Social Media and Adolescent Mental Health
Independent Research for Social Dynamics and Well-Being Lab
David Gomez
PDF | Slides

Summary: Conducted a literature review on the relationship between social media use and adolescent mental health. Divided research into correlational studies, longitudinal studies, and randomized experiments. Found that (1) heavy use of social media was consistently associated with negative mental health outcomes among adolescents, especially girls; (2) the relationship is bi-directional with the forward direction (SM -> MH) being much stronger than the reverse direction (MH -> MH) and moderated by age and gender; and (3) experiments that introduce social media consistently find adverse mental health outcomes, while experiments that remove social media depend on time-frame of mental health assessment: when mental health is assessed less than 4 weeks from removal, mental health is worse, but when assessed greater than 4 weeks from removal, mental health is improved.


Jan 2023 – Dec 2023
Linguistic Patterns of Suicide Disclosure on Social Media: A time series clustering approach
Independent Research for Social Dynamics and Well-Being Lab
David Gomez
PDF | Code

Summary: This work contributes to our understanding of broadcasting self-disclosures on social media---specifically surrounding the highly stigmatized topic of suicidality. In particular, we assess (1) whether there are any psycholinguistic patterns post-disclosure, (2) if they reflect therapeutic benefits, and (3) if we can preempt those who would benefit from such disclosures. We analyze public Twitter data of (\users) users who have disclosed some form of suicidality. We use Linguistic Inquiry and Word Count (LIWC) along with timeseries clustering to identify temporal-psycholinguistic patterns post-disclosure. We identify two clusters that are differentiated by their use of \texttt{filler} words. The majority group (73\% of users) appears to experience therapeutic benefit in the form of significantly lower usage of filler words (i.e., higher coherence) than the other group post-disclosure. We then develop a range of machine learning and deep learning classifiers that utilize only pre-disclosure information to predict whether a user would benefit from such disclosures. We achieve modest but positive results, with our best model achieving an AUC score of 0.66 over a baseline of 0.50 and a macro F1 score of 0.64 over a baseline of 0.50---indicating that there is some predictive information in the language pre-disclosure that can preempt whether someone would receive therapeutic benefit from broadcasting self-disclosures. We discuss the implications of our findings for designing new intervention strategies that can improve support provisions for those who disclose suicidality on social media.


Jan 2022 – May 2022
Affective Desensitization to Public Mass Shootings on Social Media
Tern Project: Social Computing
David Gomez, Anush Mattapalli, Jonathan Satterfield, Devon Moses
PDF | Code

Summary: WIP


May 2020 – Aug 2020
On electric propulsion thrust stands: comprehensive uncertainty analysis and modernized conceptual design
Internship Report: NASA Jet Propulsion Laboratory
David Gomez, Robert Lobbia
PDF

Summary: WIP


Jan 2020 – Dec 2020
From regression analysis to deep learning: development of improved proxy measures of state-level household gun ownership
Patterns: Cell Press (2021)
David Gomez, Zhaoyi Xu, Joseph Saleh
Project | Code | DOI | PDF

Summary: WIP


Aug 2019 – Dec 2019
Challenging the inevitability of suicide: the effect of gun regulation on overall suicide rates and portfolio of preventative measures
Term Project: Accident Causation and System Safety
David Gomez, Jose Sanchez, Danial Baum, Abdulrahman Shurayma
PDF | Poster | Slides

Summary: WIP