In recent years, the rise in social media usage both vertically in terms of the number of users by platform and horizontally in terms of the number of platforms per user has led to data explosion.

User-generated social media content provides an excellent opportunity to mine data of interest and to build resourceful applications. The rise in the number of healthcare-related social media platforms and the volume of healthcare knowledge available online in the last decade has resulted in increased social media usage for personal healthcare. In the United States, nearly ninety percent of adults, in the age group 50-75, have used social media to seek and share health information. Motivated by the growth of social media usage, this thesis focuses on healthcare-related applications, study various challenges posed by social media data, and address them through novel and effective machine learning algorithms.



The major challenges for effectively and efficiently mining social media data to build functional applications include: (1) Data reliability and acceptance: most social media data (especially in the context of healthcare-related social media) is not regulated and little has been studied on the benefits of healthcare-specific social media; (2) Data heterogeneity: social media data is generated by users with both demographic and geographic diversity; (3) Model transparency and trustworthiness: most existing machine learning models for addressing heterogeneity are considered as black box models, not many providing explanations for why they do what they do to trust them.

In response to these challenges, three main research directions have been investigated in this thesis: (1) Analyzing social media influence on healthcare: to study the real world impact of social media as a source to offer or seek support for patients with chronic health conditions; (2) Learning from task heterogeneity: to propose various models and algorithms that are adaptable to new social media platforms and robust to dynamic social media data, specifically on modeling user behaviors, identifying similar actors across platforms, and adapting black box models to a specific learning scenario; (3) Explaining heterogeneous models: to interpret predictive models in the presence of task heterogeneity. In this thesis, novel algorithms with theoretical analysis from various aspects (e.g., time complexity, convergence properties) have been proposed. The effectiveness and efficiency of the proposed algorithms is demonstrated by comparison with state-of-the-art methods and relevant case studies.

One of the critical issues in the U.S. healthcare sector is attributed to medications management. Mismanagement of medications can not only bring more unfavorable medical outcomes for patients, but also imposes avoidable medical expenditures, which can be partially accounted for the enormous $750 billion that the American healthcare system wastes annually. The lack of efficiency in medical outcomes can be due to several reasons. One of them is the problem of drug intensification: a problem associated with more aggressive management of medications and its negative consequences for patients.

To address this and many other challenges in regard to medications mismanagement, I take advantage of data-driven methodologies where a decision-making framework for identifying optimal medications management strategies will be established based on real-world data. This data-driven approach has the advantage of supporting decision-making processes by data analytics, and hence, the decision made can be validated by verifiable data. Thus, compared to merely theoretical methods, my methodology will be more applicable to patients as the ultimate beneficiaries of the healthcare system.

Based on this premise, in this dissertation I attempt to analyze and advance three streams of research that are influenced by issues involving the management of medications/treatments for different medical contexts. In particular, I will discuss (1) management of medications/treatment modalities for new-onset of diabetes after solid organ transplantations and (2) epidemic of opioid prescription and abuse.

Monitoring complex diseases and their comorbidities requires accurate and convenient measurements of multiple biomarkers. However, many state-of-the-art bioassays not only require complicated and time-consuming procedures, but also measure only one biomarker at a time. This noncomprehensive single-biomarker monitoring, as well as the cost and complexity of these bioassays advocate for a simple, rapid multi-marker sensing platform suitable for point-of-care or self-monitoring settings. To address this need, diabetes mellitus was selected as the example complex disease, with dry eye disease and cardiovascular disease as the example comorbidities. Seven vital biomarkers from these diseases were selected to investigate the platform technology: lactoferrin (Lfn), immunoglobulin E (IgE), insulin, glucose, lactate, low density lipoprotein (LDL), and high density lipoprotein (HDL). Using electrochemical techniques such as amperometry and electrochemical impedance spectroscopy (EIS), various single- and dual-marker sensing prototypes were studied. First, by focusing on the imaginary impedance of EIS, an analytical algorithm for the determination of optimal frequency and signal deconvolution was first developed. This algorithm helped overcome the challenge of signal overlapping in EIS multi-marker sensors, while providing a means to study the optimal frequency of a biomarker. The algorithm was then applied to develop various single- and dual-marker prototypes by exploring different kinds of molecular recognition elements (MRE) while studying the optimal frequencies of various biomarkers with respect to their biological properties. Throughout the exploration, 5 single-marker biosensors (glucose, lactate, insulin, IgE, and Lfn) and one dual-marker (LDL and HDL) biosensor were successfully developed. With the aid of nanoparticles and the engineering design of experiments, the zeta potential, conductivity, and molecular weight of a biomarker were found to be three example factors that contribute to a biomarker’s optimal frequency. The study platforms used in the study did not achieve dual-enzymatic marker biosensors (glucose and lactate) due to signal contamination from localized accumulation of reduced electron mediators on self-assembled monolayer. However, amperometric biosensors for glucose and lactate with disposable test strips and integrated samplers were successfully developed as a back-up solution to the multi-marker sensing platform. This work has resulted in twelve publications, five patents, and one submitted manuscripts at the time of submission.

According to sources of the Centers for Disease Control and Prevention, approximately 1.7 million traumatic brain injury (TBI) cases occur annually in the United States. TBI results in 50 thousand deaths, nearly 300 thousand hospitalizations and 2.2 million emergency room visits causing a $76 billion economic burden in direct and indirect costs. Furthermore, it is estimated that over 5 million TBI survivors in the US are struggling with long-term disabilities. And yet, a point-of-care TBI diagnostic has not replaced the non-quantitative cognitive and physiological methods used today. Presently, pupil dilation and the Glasgow Coma Scale (GCS) are clinically used to diagnose TBI. However, GSC presents difficulties in detecting subtle patient changes, oftentimes leaving mild TBI undiagnosed. Given the long-term deficits associated with TBIs, a quantitative method that enables capturing of subtle and changing TBI pathologies is of great interest to the field.

The goal of this research is to work towards a test strip and meter point-of-care technology (similar to the glucose meter) that will quantify several TBI biomarkers in a drop of whole blood simultaneously. It is generally understood that measuring only one blood biomarker may not accurately diagnose TBI, thus this work lays the foundation to develop a multi-analyte approach to detect four promising TBI biomarkers: glial fibrillary acidic protein (GFAP), neuron specific enolase (NSE), S-100β protein, and tumor necrosis factor-α (TNF-α). To achieve this, each biomarker was individually assessed and modeled using sensitive and label-free electrochemical impedance techniques first in purified, then in blood solutions using standard electrochemical electrodes. Next, the biomarkers were individually characterized using novel mesoporous carbon electrode materials to facilitate detection in blood solutions and compared to the commercial standard Nafion coating. Finally, the feasibility of measuring these biomarkers in the same sample simultaneously was explored in purified and blood solutions. This work shows that a handheld TBI blood diagnostic is feasible if the electronics can be miniaturized and large quantity production of these sensors can be achieved.