Author ORCID Identifier

https://orcid.org/0000-0003-3422-7257

Semester

Summer

Date of Graduation

2025

Document Type

Dissertation

Degree Type

PhD

College

Statler College of Engineering and Mineral Resources

Department

Industrial and Managements Systems Engineering

Committee Chair

Avishek Choudhury

Committee Member

Ashish Nimbarte

Committee Member

Imtiaz Ahmed

Committee Member

JuHyeong Ryu

Committee Member

Christopher Moore

Committee Member

Mostaan Saremi

Abstract

Mental health issues have become a significant global public health concern, especially among younger generations. The growing number of mental health challenges, combined with limited access to quality care, makes the problem even worse. Studies reveal that over 70% of individuals worldwide in need of mental health services do not receive appropriate care. Digital health technologies have the potential to enhance mental health services by making them more accessible and affordable. Despite the increasing popularity of mental health mobile applications (mHealth), there remains a lack of robust evidence of their effectiveness and the level of user trust, particularly in areas such as depression and attention deficit hyperactivity disorder (ADHD).

This dissertation aims to deepen understanding of user engagement, trust, effectiveness, and overall perception in digital mental health apps. To address this goal, four key research questions were developed. The first research question explores the effectiveness of evidence-based mental health applications through a literature review. The second question examines user acceptance and knowledge improvement resulting from the use of the mental health app. The third question focuses on quantifying users’ physiological responses when interacting with AI-generated versus doctor-generated depression screening outcomes. The fourth question focuses on quantifying objective measures to assess users’ trust in repeated decision-making tasks of AI-generated depression treatment recommendations.

We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines to address Research Question 1. To investigate Research Questions 2 and Research Questions 3, we designed and developed a web-based mental health application, My Friendly Mind, and implemented a complex mixed-methods Phase 1 clinical trial design. This experimental framework incorporated eye tracking and heart rate recordings, pre- and post-intervention surveys, and qualitative interviews. To address Research Question 4, we developed a task-based experimental module in Qualtrics focused on mental health scenarios and treatment recommendations. This intervention followed the same Phase 1 clinical trial framework to capture neurophysiological responses during trust-based decision-making.

We applied the Wilcoxon signed-rank test, Partial Least Squares Structural Equation Modeling (PLS-SEM), and thematic analysis to analyze data for Research Question 2. To address Research Question 3, we used the Mann-Whitney U test and Wilcoxon signed-rank test to analyze pupil diameter using MATLAB and employed Python-based models to predict trust using Galvanic Skin Response (GSR) signals. For Research Question 4, we conducted EEG signal analysis in MATLAB and used repeated-measures ANOVA to examine brain responses.

The novelty of this dissertation is to use objective measures to quantify the association between physiological signal activity and user trust in AI-generated depression diagnosis and treatment recommendations. The findings from this research will guide future app designs that enhance effectiveness, user acceptance, and address the nuances of user trust in AI within mental healthcare.

Download

Available for download on Tuesday, July 28, 2026

Share

COinS