Wednesday, Aug 6: 10:30 AM - 12:20 PM
4168
Contributed Posters
Music City Center
Room: CC-Hall B
Main Sponsor
Mental Health Statistics Section
Presentations
Introduction: Bipolar Disorder (BD), previously known as manic-depressive illness, is a persistent mental health condition marked by intense mood fluctuations. These mood episodes range from emotional highs (mania or hypomania) to lows (depression), significantly affecting an individual's energy, activity levels, sleep patterns, behavior, and cognitive clarity. Methods: This study utilizes both resting-state functional MRI (rs-fMRI) and microscopic imaging techniques to capture detailed structural and functional insights into brain tissue samples. Microscopy was employed to analyze cellular structures in key brain regions, providing additional context for the rs-fMRI analysis. Independent Component Analysis (ICA) was applied to rs-fMRI scans from 45 healthy controls (HCs) and 45 BD subjects to identify significant features. The top features from five selected components were then used as inputs for a three-dimensional convolutional neural network (3D-CNN) model aimed at BD diagnosis. Results: Out of 90x5=450 independent components, the model was trained on 70% (315 components) and tested on the remaining 30% (135 components). Evaluation metrics confirmed high performance in disting
Keywords
Bipolar disorder
rsfMRI
Independent Component Analysis (ICA)
Deep learning
Health risks
Co-Author(s)
Amjad Rehman, 4Artificial Intelligence & Data Analytics Lab (AIDA) CCIS Prince Sultan University
Noor Ayesha, Center of Excellence in Cyber Security (CYBEX) Prince Sultan University
Haider Ali, Department of Statistics University of Gujrat, Gujrat Pakistan
Faten Alamri, Princess Nourah Bint Abdulrahman Universty
First Author
Faten Alamri, Princess Nourah Bint Abdulrahman Universty
Presenting Author
Faten Alamri, Princess Nourah Bint Abdulrahman Universty
Inverse probability treatment weighting (IPTW) is a widely used method in causal inference to address confounding bias; however missing data frequently arises in such studies, potentially impacting the validity of causal estimates. One approach to handling missing data is to specify a missingness model and estimate the probability that an individual is a complete case and derive a corresponding missingness weight (wm). Under this approach, we use the missingness weight to estimate the treatment weight (wt), by fitting a weighted propensity score model for treatment. We conduct a simulation study to evaluate the optimal approach for incorporating missing data weights within the IPTW framework. Specifically, we compare whether using the product of missing data weights and treatment weights (wm x wt) in the final analysis model yields more accurate causal effect estimates than using treatment weights (wt) alone. Our findings will provide guidance on the optimal implementation of inverse probability weighting to address both missing data and confounding bias, ultimately strengthening the robustness of causal inference in observational research.
Keywords
Causal inference study
missing data
propensity score model
IPTW
Co-Author(s)
Manjula Tamura, Division of Nephrology, Department of Medicine, Stanford University School of Medicine, Palo Alto, C
Maria Montez-Rath, Stanford University
First Author
I-Chun Thomas, Geriatric Research, Education and Clinical Center, Veterans Affairs Palo Alto, Palo Alto, CA
Presenting Author
I-Chun Thomas, Geriatric Research, Education and Clinical Center, Veterans Affairs Palo Alto, Palo Alto, CA
Change point detection (CPD) is the process of finding distributional changes in time-ordered data streams. When CPD is applied repeatedly as data is collected, this is referred to as online CPD. Online CPD is employed in smartphone-based digital phenotyping studies in psychiatric populations to detect changes in behavior as they occur so a patient's care team can intervene to prevent a potential adverse event. By repeatedly performing CPD hypothesis tests in this way, a multiple testing problem quickly emerges. Furthermore, since consecutive tests incorporate a subset of the same days, the test statistics will be highly correlated. Thus, multiple testing methodology that assume independence or even allow for a small degree of correlation are not appropriate. In settings where we want to minimize false alarms while maintaining adequate sensitivity, it is important to use an appropriate significance level for each test that accounts for this high degree of serial correlation. We present an approach that adjusts for sequential hypothesis testing in the online CPD setting and demonstrate its effectiveness in a digital phenotyping study of patients with mood affective disorders.
Keywords
Change point detection
Sequential hypothesis testing
Digital phenotyping
Adverse Childhood Experiences (ACEs) disproportionately occur among adolescents from under-resourced communities. Further, ACEs are associated with increased prevalence of mental health disorders in adolescence, such as depression, anxiety, and posttraumatic stress. It is unclear, however, how ACEs are differentially experienced by boys and girls, and, in turn, how these experiences might impact mental health symptoms. We will use baseline data from a larger study - a randomized controlled trial of a trauma-informed mindfulness intervention for 8th graders in Baltimore City, Maryland (n=615; 74% Black, 54% female) - to analyze differences between boys' and girls' experience of both individual and cumulative ACEs. We will also conduct analyses to associate high-occuring ACEs in our sample (e.g., parental incarceration) with adolescents' depression, anxiety and posttraumatic stress symptoms, again looking at boys and girls experiences separately. This analysis will help develop a more nuanced understanding of Adverse Childhood Experiences which could help inform future interventions.
Keywords
data analysis
analysis by a high schooler
adolescent mental health
adverse childhood experiences
~ 75% of mental illness symptoms occur before age 24. Today's Gen Z and Alpha have grown up with daily exposure to digital technology. Yet widely-used mental health assessment tools like the GAD and PHQ surveys, developed over 2 decades ago, rely on episodic, simple, self-reporting methods capturing limited data about patients' lived experiences. In our increasingly digital world, new markers -- location data and screen-time patterns -- can be collected passively & continuously with minimal user burden. Analyzing this rich dataset requires sophisticated statistical approaches to handle its complexity (e.g. big data, repeated measures & and time-series data), and offers the potential for more accurate psychological assessment. We have developed an ML app that integrates passively collected data with self-reported check-ins, on over 40 million data points from > 250 students. We have identified associations between daily moods, habits, social interactions, and feelings of loneliness and acceptance, demonstrating the feasibility of novel digital biomarkers for tracking behavioral and mental health. Statisticians have the opportunity to modernize mental health tools and improve lives.
Keywords
mental health
digital biomarkers
ML/AI modeling of big data
According to the World Health Organization about 4% of adult men and about 6% of adult women suffer from depression globally. In this study, we used a subset of Investigating Gains in Neurocognition in an Intervention Trial of Exercise (IGNITE) dataset. There were 648 participants between ages 65 and 80 in IGNITE study. We used data collected from multiple instruments. We will present the associations between a measurement of depression and other measurements representing individual characteristics. Hospital Anxiety and Depression Scale (HADS) and Geriatric Depression Scale (GDS) were used to measure the individual depression level. Multiple regression models adjusted for demographic variables were utilized to test for associations. Socio-economic variables considered had correlations below 0.14 in magnitude with both depression scales. Results of multiple regression with several other variables using standard instruments (Loneliness, satisfaction with life, stress, physical and mental health, and emotional intelligence) will be discussed.
Keywords
Depression
Socio-Economic Variables
Mental and Social Health
This study explores the utility of Sliced Inverse Regression (SIR) for dimension reduction in analyzing a multivariate endogenous variable related to mental health outcomes. Understanding the complex interdependencies between mental health and a wide array of covariates-spanning demographic, biological, genetic, and physiological domains-presents a significant statistical challenge. Traditional regression models often struggle with multicollinearity and high-dimensional data, limiting their ability to uncover meaningful relationships. By applying SIR, we reduce the dimensionality of the covariate space while preserving key directions that explain variation in the mental health outcome. Our analysis identifies the most influential covariate directions and reveals interpretable subspaces that capture underlying mental health dynamics. Results suggest that combinations of genetic markers, age, socioeconomic status, and physiological metrics play significant roles in mental health variability. The findings highlight SIR's potential to uncover complex nonlinear associations and its value in guiding further research on personalized interventions and targeted mental health treatments.
Keywords
Sliced Inverse Regression
Mental Health
Dimensionality Reduction
First Author
Jonathan Day, United States Military Academy
Presenting Author
Jonathan Day, United States Military Academy