MetadataShow full item record
PublisherThe University of Arizona.
RightsCopyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction, presentation (such as public display or performance) of protected items is prohibited except with permission of the author.
AbstractValue-based healthcare is an emerging healthcare delivery model which incentivizes and rewards physicians for improved patient outcomes and quality of care, rather than the amount of services. The objective of value-based healthcare is to move the healthcare delivery system from reactive disease management to proactive care. Such proactive healthcare necessitates an active assessment of multi-dimensional health risks. Major health risks, such as medication nonadherence, disease risk, and hospital readmission, cost the US healthcare systems over $300 billion per year and escalate morbidity and mortality risks. Early detection of those health risks could significantly benefit patients and the health sectors from both disease management and financial perspectives. My dissertation focuses on health risk analytics that serves as the analytical foundation for proactive health risk assessment. This dissertation designs fine-grained deep learning methods leveraging big data, including health social media and clinical claims, to provide analytical capabilities for several critically important health risks: vaping, medication nonadherence, opioid addiction, and hospital readmission. This dissertation presents four essays to tackle these health risks. The first essay devises a deep learning model to identify health risks of vaping. The second essay develops a sentiment-enriched deep learning method to understand patient medication nonadherence from health social media. The third essay presents a multi-view deep learning approach to discover the treatment barriers of opioid addiction. The fourth essay designs a novel deep learning framework that incorporates an illness trajectory to predict hospital readmission risk. The presented frameworks, systems, and design principles contribute to computational data science, deep learning, and predictive modeling research domains.
Degree ProgramGraduate College
Management Information Systems