LEVERAGING PASSIVELY-COLLECTED WEARABLE ACCELEROMETRY DATA COUPLED WITH MACHINE LEARNING TO LONGITUDINALLY DETECT AND PREDICT MAJOR DEPRESSIVE DISORDER - A Thesis

Aug 5, 2026

Upload by Maeve

Major depressive disorder (MDD) is a debilitating, heterogeneous mental health condition defined by symptoms such as low mood, sleep problems, psychomotor difficulties, and fatigue. Given that MDD affects roughly one in twenty adults globally and has increased in the United States, effective screening, diagnosis, and treatment remain urgent. This thesis investigates whether longitudinal, passively collected wearable accelerometry can detect and predict MDD-related outcomes. Using nationally representative and clinical samples, unsupervised, supervised, and deep learning models characterize movement, sedentary, and sleep behaviors to model depression presence, symptom severity, and variability. The work supports real-world clinical potential.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages209

Reading Time8 h 47 min

AccessFree to download

Abstract

Preface

Methods and Data Sources

Machine Learning and Deep Learning Approaches

Outcomes: Detection and Prediction of MDD

Dartmouth College Dartmouth Digital Commons | Dartmouth College Ph. D Dissertations | Theses and Dissertations | | --- | --- | | Spring 3-5-2024
LEVERAGING PASSIVELY-COLLECTED WEARABLE ACCELEROMETRY DATA COUPLED WITH MACHINE LEARNING TO LONGITUDINALLY DETECT AND PREDICT MAJOR DEPRESSIVE DISORDER
George D. Price
Dartmouth College, [george.price.gr@dartmouth.edu](george.price.gr@dartmouth.edu)
Follow this and additional works at: [https://digitalcommons.dartmouth.edu/dissertations](https://digitalcommons.dartmouth.edu/dissertations)
Part of the Quantitative Psychology Commons | | Recommended Citation Price, George D., "LEVERAGING PASSIVELY-COLLECTED WEARABLE ACCELEROMETRY DATA COUPLED WITH MACHINE LEARNING TO LONGITUDINALLY DETECT AND PREDICT MAJOR DEPRESSIVE DISORDER" (2024) . Dartmouth College Ph. D Dissertations. 279. [https://digitalcommons.dartmouth.edu/dissertations/279](https://digitalcommons.dartmouth.edu/dissertations/279) This Thesis (Ph. D.) is brought to you for free and open access by the Theses and Dissertations at Dartmouth Digital Commons. It has been accepted for inclusion in Dartmouth College Ph. D Dissertations by an authorized administrator of Dartmouth Digital Commons. For more information, please contact [dartmouthdigitalcommons@groups.dartmouth.edu](dartmouthdigitalcommons@groups.dartmouth.edu). LEVERAGING PASSIVELY-COLLECTED WEARABLE ACCELEROMETRY DATA COUPLED WITH MACHINE LEARNING TO LONGITUDINALLY DETECT AND PREDICT MAJOR DEPRESSIVE DISORDER A Thesis Submitted to the Faculty In partial fulfillment of the requirements for the Degree of Doctor of Philosophy in Quantitative Biomedical Sciences by George D. Price Guarini School of Graduate and Advanced Studies Dartmouth College Hanover, New Hampshire March 2024 Examining Committee: | (chair) Nicholas Jacobson | | --- | | Jennifer Emond | | Lisa Marsch | Varun Mishra F. Jon Kull, Ph.D. Dean of the Guarini School of Graduate and Advanced Studies ABSTRACT Major depressive disorder (MDD) is a debilitating and heterogenous mental health disorder that is characterized by symptoms including low mood, issues with sleep, psychomotor difficulties, and fatigue. MDD affects one in twenty adults worldwide and has shown increased prevalence in the United States over the past twenty years. As such, efforts to effectively screen, diagnose, and treat MDD are paramount. However, to address these concerns, an increased understanding of an individual’s daily behavior is required, which is not adequately captured by infrequent clinical visits. One such method for consistent, longitudinal observation, passively-collected accelerometry, serves as an observational method for unobtrusively capturing movement, sedentary and sleep behaviors in real-time. Therefore, the primary effort of this thesis work is to investigate the utility of leveraging longitudinal, passively-collected accelerometer information in detecting and predicting outcomes related to MDD. The primary data source for this work comes from the nationally representative 2011-2014 National Health And Nutrition Examination Survey and two separate clinical samples. A combination of unsupervised machine learning, supervised machine learning, and deep learning techniques were used to characterize movement, sedentary and sleep behaviors for individuals with MDD, as well as investigate depression presence, individual depressive symptoms, long-term depression variability, and acute depression variability. Taken together, this work highlights the utility of using solely passively-collected accelerometry to investigate outcomes related to MDD and seeks to provide insight into the utility of implementing such approaches in real-world clinical settings to both improve our understanding of, and opportunities for, clinical engagement for individuals with MDD. PREFACE I was first introduced to academic research during my sophomore year at Boston College when I j

FAQs

What is the main goal of this thesis?

To evaluate how longitudinal, passively collected wearable accelerometry data can detect and predict outcomes related to major depressive disorder (MDD).

What data sources are used for the study?

The primary data come from the nationally representative 2011–2014 National Health And Nutrition Examination Survey, supplemented by two separate clinical samples.

Which outcomes related to MDD are modeled?

The thesis models depression presence, individual depressive symptoms, long-term depression variability, and acute depression variability.

Aug 5, 2026

Upload by Maeve