Research
Time-Series Prediction of Change in Depressive Symptoms
Constructed an ML pipeline predicting changes in depression severity in adolescents over three months. Trained Elastic Net, Random Forest, and XGBoost models on time-series features from ecological momentary assessments. Computed residuals from baseline predictions to capture unexplained change factors.
Structural Stigma and Digital Intervention
Investigated whether structural stigma and social support moderated treatment responses among LGBTQ+ youth receiving a digital single-session intervention. Created a county-level structural stigma factor using confirmatory factor analysis across 181 counties in 46 U.S. states.
Digital Interventions for Rural Adolescents
Evaluated the viability of social media-based recruitment of rural adolescents for online single-session interventions. Analyzed data from 2,322 adolescents in a web-based RCT, finding comparable completion rates and depression symptom reductions between rural and urban youth.
Childhood Trauma and Onset of Suicidal Attempt
Examined the relationship between childhood traumatic experiences and early/late-onset suicidal behavior among 224 depressed older adults. Used Gaussian mixture modeling to identify a cutoff age of 30 for early vs. late-onset attempters.
Suicidal Ideation Trajectories — A Latent Profile Analysis
Used Latent Profile Analysis in a cohort of depressed older adults to identify distinct ideation profiles, their clinical correlates, and association with risk of suicidal behavior longitudinally. Depicted clinical characteristics using radar plots and time-series data spanning 2500 days.
Structural Stigma and Suicidal Thoughts and Behavior
Examined associations between state-level structural stigma and suicidal thoughts and behaviors in a nationwide sample of sexual minority adolescents (n=489). Found that state-level structural stigma was positively associated with higher levels of lifetime suicidal ideation.
Computational Psychiatry — LASSO Prediction of Self-Injury
Built a supervised machine learning model using LASSO regularized logistic regression with 31 baseline characteristics to predict non-suicidal self-injury among gender diverse individuals. Achieved AUC of 0.85 with 10-fold cross-validation.
Transgender Identity Development and Minority Stress
Studied the role of childhood gender nonconformity in developmental milestones of TGNB identity development. Found that TGNB individuals who were nonconforming in childhood reached identity development milestones earlier and reported lower levels of felt stigma and depression.
Health Inequalities and Suicide Risk — A Systematic Review
Systematic review assessing the psychological effects of gender-affirming hormones and surgical treatments on gender diverse children, adolescents, and young adults. Sponsored by International Academy of Suicide Research.