Investigations of Factors Associated With Treatment Response and Prognosis in Anxiety Disorders
Investigations of Factors Associated With Treatment Response and Prognosis in Anxiety Disorders
The purpose of this study is to build a comprehensive brain and behavioral dataset on anxiety disorders and identify clinical factors related to treatment response and prognosis in patients with anxiety disorders. By conducting an integrated analysis of clinical features (phenotypes), comorbid conditions, virtual reality (VR)-based behavioral data, electroencephalogram (EEG) data, and MRI-based data in patients undergoing treatment for anxiety disorders and healthy controls, this study aims to enhance the understanding of the pathophysiology of anxiety disorders and establish a foundation for precision medicine.
Anxiety disorders significantly impair quality of life by inducing irrational fear and apprehension even in everyday situations. Over the past few years, the prevalence of anxiety disorders and associated medical expenses have surged, escalating the societal and economic burden. However, the current diagnostic system, DSM-5, relies on symptom-based approaches, which often lead to heterogeneity within the same diagnostic category. This limitation hinders the accurate prediction of treatment response and the design of optimal therapeutic strategies. To address these issues, the U.S. National Institute of Mental Health (NIMH) proposed the Research Domain Criteria (RDoC). RDoC is a multidimensional framework that integrates behavioral, psychological, and neurological data to facilitate the development of treatment strategies rooted in pathophysiology. Nonetheless, the acquisition of objective and reliable assessment data remains a challenge due to the inherent characteristics of mental disorders.
Virtual reality (VR) technology has emerged as a promising solution to overcome these limitations. VR enables the creation of immersive 3D environments that can elicit anxiety-related behaviors and precisely evaluate them in controlled settings. It also allows data collection in scenarios that closely resemble real-life situations while offering advantages such as repeatability and ease of use. Combining behavioral data collected in VR environments with EEG data can provide a more detailed understanding of the neurophysiological representations underlying anxiety behaviors. EEG, as a biomarker reflecting neural activity in real time, facilitates intuitive and effective analyses. Moreover, MRI-based brain imaging plays a critical role in identifying structural and functional connectivity differences between anxiety disorder patients and healthy controls. It offers crucial insights into the biomarkers that predict treatment outcomes and prognosis.
By integrating these data, this study seeks to uncover the key factors influencing treatment response and prognosis in patients with anxiety disorders.
[ Observational Parameters, Clinical and Laboratory Assessments ]
Participants in the anxiety behavior group will undergo four visits (initial evaluation, 2-month follow-up, 6-month follow-up, and 12-month follow-up), while the healthy control group will attend two visits (initial evaluation and 12-month follow-up). As the evaluations at 2 and 12 months consist solely of basic clinical assessments, they may be conducted via remote methods, such as telephone interviews or online surveys (e.g., Google Forms).
Initial Evaluation:
The initial evaluation includes approximately one hour of clinical and psychiatric symptom evaluations, a 30-minute virtual reality-based behavioral assessment, and a 40-minute brain imaging session.
Clinical assessment:
Includes the assessment of sociodemographic factors (e.g., gender, age, education level, marital status, occupation, alcohol and smoking habits, height, and weight) and anxiety disorder-related medical history (e.g., symptom patterns and duration, psychiatric treatment history, and family history).
Evaluation of anxiety symptoms and associated psychiatric conditions:
Utilizes various standardized scales and questionnaires, such as PDSS, LSAS, GAD, DASS, IUS, ASI, BAS, PANAS, neurocognitive tests, SSQ, and STRAIN.
Virtual Reality-Based Behavioral Assessment:
Gathers biometric signals and EEG data in response to stimuli presented in virtual environments.
Brain MRI:
Conducted using a 3.0T Philips Ingenia CX scanner, acquiring T1-weighted 3D coronal structural images, resting-state functional MRI, and diffusion-weighted images.
Follow-Up Assessments:
Participants in the anxiety behavior group will undergo assessments of four measures - PDSS, LSAS, GAD, and DASS - at 2 and 12 months. At 6 months, they will undergo assessments covering (2) the evaluation of anxiety symptoms and associated psychiatric characteristics, (3) virtual reality-based behavioral assessments, and (4) brain MRI, as conducted at baseline.
Healthy control participants will undergo the same four assessments (PDSS, LSAS, GAD, DASS) evaluating anxiety symptoms and associated psychiatric characteristics at 12 months at the Department of Psychiatry of our institution.
At 12 months, both the anxiety behavior group and the healthy control group will additionally complete the collection of clinical evaluation items (1), excluding sex and age.
[ Data Analysis and Statistical Methods ] The results of the initial evaluation will be used to develop a three-group classification model (healthy controls vs. response vs. no-response) based on clinical evaluation outcomes at 2, 6, and 12 months. Participants will be categorized into remission and no-remission groups based on a ≥40% reduction in the PDSS score at the 2, 6, and 12 months follow-up assessments, allowing for a detailed understanding of the treatment effects over time.
Additionally, a regression model will be constructed to analyze the magnitude of changes in clinical evaluation outcomes.
This study will test various machine learning models, including classification methods like Dummy, Logistic Regression, Support Vector Machine, and Boosting models (eXtreme Gradient Boosting, CatBoost), using k-fold cross-validation to evaluate performance through accuracy, balanced accuracy, and F1 scores. Regression models, including Dummy Regressor, Linear Regression, Support Vector Regression, Multilayer Perceptron, and Bayesian Regression, will be assessed using Mean Squared Error (MSE), Mean Absolute Error (MAE), and R² scores.
The pathophysiology of anxiety disorder will be explored using resting-state functional brain MRI data, with cross-sectional and longitudinal comparisons of anxiety behavior groups and healthy controls. Network analysis will be conducted to identify significant brain connections, with false discovery rate (FDR) correction applied to control for multiple comparisons.
Inclusion Criteria
Anxiety Behavior Group
Healthy Control Group
Exclusion Criteria:
Anxiety Behavior Group
Healthy Control Group
jihndy.kim@samsung.com+82-2-2001-2431
+82-2-2001-1805