CLINICAL AND SOCIAL PATHWAYS TO CARE: A COMPUTATIONAL EXAMINATION OF SOCIAL MEDIA FOR MENTAL HEALTH CARE
Author(s)
Ernala, Sindhu Kiranmai
Advisor(s)
Editor(s)
Collections
Supplementary to:
Permanent Link
Abstract
In the last decade, powered by connectivity to large social networks and advances in collecting and analyzing digital traces of individuals from social media platforms, researchers have gleaned rich insights into individuals’ and populations’ mental health states and experiences, including their moods, emotions, social interactions, language, and communication patterns. Using these inferences, researchers have been able to study support-seeking behaviors, distinguishing patterns, risk markers, and diagnosis states for mental illnesses from social media data, promising a fundamental change in mental health care. What we need next in this line of work is for data and algorithms based on social media to be contextualized in people’s pathways to mental health care. However, there are several challenges and unanswered questions that present hurdles.
First, gaps exist in the psychometric validity of social media based measurements of behaviors and the utility of these inferences in predicting clinical outcomes in patient populations. Second, if social media can act as an intervention platform, outside of discrete events, a holistic understanding of its role in people’s lives along the course of a mental illness is crucial. Lastly, several questions remain around the ethical implications of research practices in engaging with a vulnerable population subject to this research.
This thesis charts out empirical and critical understandings and develops novel computational techniques to ethically and holistically examine how social media can be employed to support mental health care. Focusing on schizophrenia, one of the most debilitating and stigmatizing of mental illnesses, this thesis contributes a deeper understanding on pathways to care via social media along three themes: 1) prediction of clinical mental health states from social media data to support clinical interventions, 2) understanding online self-disclosure and social support as pathways to social care, and 3) the intersection of social and clinical pathways to care along the course of mental illness. In doing so, this work combines theories from social psychology, computer-mediated communication, and clinical literature with machine learning, statistical modeling, and natural language analysis methods applied on large-scale behavioral data from social media platforms. Together, this work contributes novel methodologies and human-centered algorithmic design frameworks to understand the efficacy of social media as a mental health intervention platform, informing clinicians, researchers, and designers who engage in developing and deploying interventions for mental health and well-being.
Sponsor
Date
2021-07-29
Extent
Resource Type
Text
Resource Subtype
Dissertation