Date of Completion

7-14-2020

Embargo Period

7-14-2021

Keywords

Ubiquitous computing, machine learning, depression prediction, sensor data analytics

Major Advisor

Dr. Bing Wang

Associate Advisor

Dr. Jinbo Bi

Associate Advisor

Dr. Alexander Russell

Field of Study

Computer Science and Engineering

Degree

Doctor of Philosophy

Open Access

Open Access

Abstract

Depression is a common, yet serious health problem. It has significant detrimental impacts on both physical and psychological functioning. Current diagnosis techniques rely on physician-administered or patient self-administered interview tools, which are burdensome and suffer from recall bias. Additionally, these techniques incur higher medical costs. There is an urgent need for an accurate, objective and easily accessible depression screening tool for mass usage. In this dissertation, we explore the usage of smartphone sensing data, collected directly on smartphones or meta-data collected from a WiFi infrastructure, for automatic depression screening and depressive symptom prediction.

In the first part of the dissertation, we develop a novel approach that investigates the feasibility of automatic large-scale depression prediction using meta-data captured in an institution’s WiFi network, without direct data capture (i.e., running apps) on phones. Specifically, when smartphones connect to a WiFi network, their locations (and hence the locations of the users) can be determined by the access points that they associate with; the location information over time provides important insights into the behavior of the users, which can be used for depression screening. To investigate the feasibility of this approach, we have analyzed two datasets, each collected over several months, involving tens of participants recruited from a university. Our results demonstrate that WiFi meta-data is effective for passive depression screening: the F1 scores are as high as 0.85 for predicting depression, comparable to those obtained by using sensing data collected directly from smartphones.

In the second part of the dissertation, we explore the feasibility of using smartphone sensing data for automatic prediction of all major categories of depressive symptoms, including both cognitive (in interests, mood, concentration) and behavioral (in appetite, energy level, sleep) symptoms. Specifically, we consider two types of smartphone data, one collected passively on smartphones and the other collected from an institution’s WiFi infrastructure and construct a family of machine learning based models for the prediction. Both scenarios require no efforts from the users and can provide objective assessment of depressive symptoms. Our results demonstrate that smartphone data can be used to predict both behavioral and cognitive symptoms effectively, with F1 score as high as 0.86. Our study makes a significant step forward over existing studies, which only focus on predicting the overall depression status (i.e., whether one is depressed or not).

In the third part of the dissertation, we explore the impact and influence of social interaction related features on mental health and wellness. Specifically, using a family of machine learning models, we have used Short Message Service (SMS) and call record data for depression prediction. By using the social interaction data collected via an Android phone app from college age students. Our results demonstrate that social interaction data can be used to predict depression effectively, with F1 score as high as 0.80. Our results show that the people with depression spend more time on the phone calls than the non-depressed population. We also found that majority of the depressed individuals send and receive fewer number of messages and communicate with a limited number of contacts.

COinS