Wednesday, Aug 6: 8:30 AM - 10:20 AM
4148
Contributed Papers
Music City Center
Room: CC-207A
Main Sponsor
Mental Health Statistics Section
Presentations
Multinomial regression is a powerful tool for modeling categorical outcomes of two or more classes. However, several challenges include information loss from categorization, & increased complexity from multiple linear models, leading to parameter inflation. Existing variable selection techniques improve model sparsity but struggle with more sparser data, often missing true signals & introduce false positives. L0-norm regularization induces exact sparsity, but is computationally prohibitive due to its non-convex, NP-hard nature. Existing software for L0 is slow, & higher data complexity worsens inefficiency. To address these challenges, we propose an L0L2 multinomial logistic regression algorithm enabling precise feature selection while maintaining computational feasibility. Our approach integrates a systematic swapping mechanism to enhance optimization & employs Iterative Reweighted Least Squares (IRLS) to enhance efficiency. This proposal is highly motivated by our real-world genetic dataset, consisting of several hundred SNP predictors associated with multi-category mental health outcomes exposed to traumatic events, mental health disorders, & substance use disorders.
Keywords
high dimensional, multinomial
sparsity, feature selection
optimization
efficiency
IRLS
L0L2 regularization
Every year, different researchers update randomized response models to improve model efficiency. In moving towards a practical application of this model, and in consideration of this year's JSM 2025 theme of "Statistics and Data Science, an AI Enriching Society" one can consider decisions with substance-abuse mental health protocol changes with combining randomized response statistical techniques, along with current scientific advances in neuroscience. This includes research regarding stress pathways and amygdala activation, which affect the response given by the participant in survey research with sensitive data. Defining terms such as emotional physiology and internal subconscious anxiety allows a researcher to restructure the randomized response model for practicality over efficiency. We will examine the 2-deck of cards model, proposed by Odumade and Singh (2009), and show the adjustments needed for practicality of a real world application highlighting vaping via e-cigarette use amongst teens during the COVID-19 pandemic.
Keywords
survey sampling
randomized response
sensitive data
mental health
substance abuse
public health
Depression and anxiety are debilitating and prevalent diagnoses with wide-reaching negative psychological and economic impacts. Clinicians note that depression and anxiety, although distinct conditions, often occur together in patients, with little information explaining such comorbidity. In absence of information on the underlying aetiology of these diseases, some clinicians hypothesize that one trait may predispose another, thereby inducing a direction of dependence between these psychological traits.
The Intern Health Study (IHS) examines self-reported depression and anxiety among doctors in residency programs in the US. Being able to establish a sense of directionality between anxiety and depression to understand the dominance between these two mental health outcomes is critical to develop adequate clinical diagnostics and administer medical intervention. We propose a novel information-theoretic coefficient that leverages Shannon's entropy metric used to examine directed dependence between anxiety and depression. The proposed method is evaluated by simulation studies and applied to IHS data, where a dominating effect of depression on anxiety is observed in medical interns.
Keywords
bivariate causal discovery
information theory
mental health outcomes
Major depressive disorder (MDD), a leading cause of disability and mortality, is associated with reward-processing abnormalities and concentration issues. Motivated by the probabilistic reward task from the EMBARC study, we propose a novel framework that integrates the reinforcement learning (RL) model, hidden Markov model (HMM), and drift-diffusion model (DDM) to analyze reward-based decision-making alongside response times. To model latent state switching, we use an HMM. In the 'engaged' state, decisions follow an RL-DDM, simultaneously capturing reward processing, decision dynamics, and temporal structure. In contrast, in the 'lapse' state, decision-making is modeled using a simplified DDM, where specific parameters are fixed to approximate random guess. The method is implemented via a computationally efficient EM algorithm with forward-backward procedures. Numerical studies show superior performance over competing methods across various settings. When applied to EMBARC, our framework reveals that MDD patients engage less than healthy controls and take longer to decide when engaged. We also examine associations between brain activities and decision-making characteristics.
Keywords
Brain–behavior association
Cognitive modeling
Drift-diffusion models
Mental health
Reinforcement learning
State switching
Suicidal ideation is a pressing mental health concern, particularly among individuals with posttraumatic stress disorder (PTSD). Early detection and timely intervention are critical yet challenging in psychiatric care. Ecological momentary assessment (EMA) via mobile devices (e.g., smartphones, wearables) provides a novel approach for continuous monitoring of psychological, behavioral, and contextual biomarkers associated with suicide risk. However, the intensive longitudinal nature of EMA data presents statistical challenges alongside opportunities for new medical insights. This study utilized generalized linear mixed-effects models to explore the relationship between coping plan use frequency and suicidal ideation, addressing both within-person and between-person variability. Significant associations were observed at both levels, with moderation analyses revealing that the relationship varied by coping strategy (CRP versus SP).
Our findings highlight the statistical complexities of EMA data and the value of tailored modeling approaches in capturing the dynamic interplay between coping behaviors and suicide risk, offering critical insights for clinical intervention.
Keywords
Ecological momentary assessment (EMA)
mHealth (Mobile Health)
Generalized linear mixed-effects models (GLMM)
Suicidal ideation
Intensive longitudinal data
Co-Author(s)
Xiaoxuan Cai, The Ohio State University
Jiaxin Chen, The Ohio State University
Samantha Daruwala, The Ohio State University Wexner Medical Center
Lauren Khazem, The Ohio State University Wexner Medical Center
Heather Wastler, The Ohio State University Wexner Medical Center
Nicholas Allan, The Ohio State University Wexner Medical Center
AnnaBelle Bryan, The Ohio State University Wexner Medical Center
Craig Bryan, The Ohio State University Wexner Medical Center
First Author
Melanie Bozzay, The Ohio State University Wexner Medical Center
Presenting Author
Jiaxin Chen, The Ohio State University
Multivariate bounded discrete data arises in many fields. In the setting of dementia studies, such data is collected when individuals complete neuropsychological tests. We outline a modeling and inference procedure that can model the joint distribution conditional on baseline covariates, leveraging previous work on mixtures of experts and latent class models. Furthermore, we illustrate how the work can be extended when the outcome data is missing at random using a nested EM algorithm. The proposed model can incorporate covariate information and perform imputation and clustering. We apply our model on simulated data and an Alzheimer's disease data set.
Keywords
Mixture models
Multivariate discrete data
Latent variable models
Binomial product mixture
Missing data
A key step in the analysis of longitudinal changes of polysubstance cravings in MOUD treatment using linear mixed model (LMM) is to choose a suitable covariance structure. Data was selected from the National Drug Abuse Treatment Clinical Trials Network protocol-0051. Opioid-dependent participants were randomly assigned to receive BUP-NX (n=287) or XR-NTX (n=283). Measures of opiate, alcohol, stimulant, and nicotine craving were collected at baseline and every 4 weeks for 8 months. Both AIC and BIC statistics revealed that the unstructured (UN) covariance structure is the best from ten common covariance structures for these four craving measures. Using a UN model in the LMM, all four craving measures declined rapidly from baseline. Baseline depression and alcohol, amphetamine, and cocaine use disorders were associated with an increased risk of alcohol, stimulant and nicotine cravings. In conclusion, the UN covariance structure is the best in the LMM, while screening for depression and multiple substances may help clinicians identify patients at a higher vulnerability for opioid relapse secondary to possible increased cravings of substances.
Keywords
Opioid use disorders
Clinical trial
Linear mixed model
Model selection
Substance use
Craving