Advancements in high-throughput technologies have led to an exponential increase in the generation of multi-modal data in computational biology. These datasets, comprising diverse biological measurements such as genomics, transcriptomics, proteomics, metabolomics, and imaging data, offer a comprehensive view of biological systems at various levels of complexity. However, integrating and analyzing such heterogeneous data present significant challenges due to differences in data modalities, scales, and noise levels. Another challenge for multi-modal analysis is the complex interaction network that the modalities share. Understanding the intricate interplay between different biological modalities is essential for unraveling the underlying mechanisms of complex biological processes, including disease pathogenesis, drug response, and cellular function. Machine learning algorithms have emerged as indispensable tools for studying multi-modal data in computational biology, enabling researchers to extract meaningful insights, identify biomarkers, and predict biological outcomes. In this dissertation, we first propose a multi-modal integration framework that takes two interconnected data modalities and their interaction network to iteratively update the modalities into new representations with better disease outcome predictive abilities. The deep learning-based model underscores the importance and performance gains achieved through the incorporation of network information into integration process. Additionally, a multi-modal framework is developed to estimate protein expression from mRNA and microRNA (miRNA) expressions, along with the mRNA-miRNA interaction network. The proposed network propagation model simulates in-vivo miRNA regulation on mRNA translation, offering a cost-effective alternative to experimental protein quantification. Analysis reveals that predicted protein expression exhibits a stronger correlation with ground truth protein expression compared to mRNA expression. Moreover, the effectiveness of integrative models is contingent upon the quality of input data modalities and the completeness of interaction networks, with missing values and network noise adversely affecting downstream tasks. To address these challenges, two multi-modal imputation models are proposed, facilitating the imputation of missing values in time series data. The first model allows the imputation of missing values in time series gene expression utilizing single nucleotide polymorphism (SNP) data for children at high risk of type 1 diabetes. The imputed gene expression allows us to predict the progression towards type 1 diabetes at birth with six years prediction horizon. Subsequently, a follow-up study introduces a generalized multi-modal imputation framework capable of imputing missing values in time series data using either another time series or cross-sectional data collected from the same set of samples. These models excel at imputation tasks, whether values are missing randomly or an entire time step in the series is absent. Additionally, leveraging the additional modality, they are able to estimate a completely missing time series without prior values. Finally, to mitigate noise in the interaction network, a link prediction framework for drug-target interaction prediction is developed. This study demonstrates exceptional performance in cold start predictions and investigates the efficacy of large language models for such predictions. Through a comprehensive review and evaluation of state-of-the-art algorithms, this dissertation aims to provide researchers with valuable insights, methodologies, and tools for harnessing the rich information embedded within multi-modal biological datasets.

This study examines the technology and appointment scheduling habits of older adults over the age of 55 through an exploratory sequential three phase mixed methods study. Phase One of this study examined features of ten existing mHealth applications through a qualitative content analysis, then a mHealth wireframe was developed from the app to replicate in addition to a redesigned version. Phase Two of the study was a thirty-four questions survey with 40 participants that inquired about their background with appointment scheduling, prior experience with technology, and demographics. After that, the mHealth applications were revised into two mHealth application prototypes. Lastly, Phase Three conducted a user test with the two mHealth prototypes through A/B testing with 15 participants. Findings from the survey showed the preferred method of scheduling an appointment among participants was primarily in-person or by phone. The user test revealed that some participants were willing to use a mHealth application to schedule an appointment if it was deemed easy to use. Recommendations for future research suggests that the iterative design process of a prototype with an underserved population would garner feedback inclusive of those older adults who are less tech savvy. The major contribution of this research was the development of the mHapps Framework which will be tested in a future study.

This dissertation presents a comprehensive exploration and implementation of attention mechanisms and transformers on several healthcare-related and assistive applications. The overarching goal is to demonstrate successful implementation of the state-of-the-art approaches and provide validated models with their superior performance to inform future research and development. In Chapter 1, attention mechanisms are harnessed for the fine-grained classification of white blood cells (WBCs), showcasing their efficacy in medical diagnostics. The proposed multi-attention framework ensures accurate WBC subtype classification by capturing discriminative features from various layers, leading to superior performance compared to other existing approaches used in previous work. More importantly, the attention-based method showed consistently better results than without attention in all three backbone architectures tested (ResNet, XceptionNet and Efficient- Net). Chapter 2 introduces a self-supervised framework leveraging vision transformers for object detection, semantic and custom algorithms for collision prediction in application to assistive technology for visually impaired. In addition, Multimodal sensory feedback system was designed and fabricated to convey environmental information and potential collisions to the user for real-time navigation and grasping assistance. Chapter 3 presents implementation of transformer-based method for operation-relevant human activity recognition (HAR) and demonstrated its performance over other deep learning model, long-short term memory (LSTM). In addition, feature engineering was used (principal component analysis) to extract most discriminatory and representative motion features from the instrumented sensors, indicating that the joint angle features are more important than body segment orientations. Further, identification of a minimal number and placement of wearable sensors for use in real-world data collections and activity recognitions, addressing the critical gap found in the respective field to enhance the practicality and utility of wearable sensors for HAR. The premise and efficacy of attention-based mechanisms and transformers was confirmed through its demonstrated performance in classification accuracy as compared to LSTM. These research outcomes from three distinct applications of attention-based mechanisms and trans- formers and demonstrated performance over existing models and methods support their utility and applicability across various biomedical and human activity research fields. By sharing the custom designed model architectures, implementation methods, and resulting classification performance has direct impact in the related field by allowing direct adoption and implementation of the developed methods.

<p>Chronic low back pain is the second most prevalent chronic condition in Canadian primary care settings. The treatment and diagnosis of chronic pain is challenging for primary care clinicians. Their main challenges are lack of knowledge and their approach toward assessing and treating pain. Evidence based guidelines have been developed for neuropathic pain and low back pain.</p> <p>CDSSs for chronic diseases are becoming popular in primary care settings as a mean to implement CPGs. A CDSS prototype for diagnosis and treatment of chronic, non-cancer pain in primary care was developed at McMaster University. It is evident that poor usability can hinder the uptake of health information technologies.</p> <p>The objective of this study was to test the usability of Pain Assistant using think aloud protocols with SUS scores in 2 iterations. In this study 13 primary care providers including family physicians, nurse practitioners and residents used Pain Assistant to complete 3 different patient case scenarios. Participants were asked to comment on both barriers and facilitators of usability of Pain Assistant. Additionally time to complete patient case scenarios was calculated for each participant. A comparison questionnaire gathered user preference between introducing CPGs in paper format and computerized decision support system.</p> <p>This study showed that iterative usability testing of the Pain Assistant with participation of real-end users has the potential to uncover usability issues of the Pain Assistant. Problems of user interface were the main usability barrier in first testing iteration following by problems of content. Changes were made to system design for second round based on the issues came up in the first iteration. However, because of time constrains not all the changes were implemented for second round of testing. Most of the refinements were to resolve user interface issues. In the second iteration, the problems with the content of Pain Assistant were the major barrier. The changes to the system design were successful in resolving user interface problems since the changed issues did not come up again in second round. Pain Assistant had an above the average usability score however no significant changes seen in SUS score. The time needed to complete tasks remained identical in both iterations. In addition, participants preferred to have CPGs in electronic formats than paper. Further study after implementing all the system changes needed to determine the effectiveness of system refinements.</p> / Master of Science (MSc)

The current dissertation provides an examination of health information technology (HIT) by analyzing big datasets. It contains two separate essays focused on: (1) the evolving intellectual structure of the healthcare informatics (HI) and healthcare IT (HIT) scholarly communities, and (2) the impact of social support exchange embedded in social interactions on health promotion outcomes associated with online health community use. Overall, this dissertation extends current theories by applying a unique combination of methods (natural language processing, machine learning, social network analysis, and structural equation modeling etc.) to the analyses of primary datasets. The goal of the first study is to obtain a full understanding of the underlying dynamics of the intellectual structures of HI and its sub-discipline HIT. Using multiple statistical methods including citation and co-citation analysis, social network analysis (SNA), and latent semantic analysis (LSA), this essay shows how HIT research has emerged in IS journals and distinguished itself from the larger HI context. The research themes, intellectual leadership, cohesion of these themes and networks of researchers, and journal presence revealed in our longitudinal intellectual structure analyses foretell how, in particular, these HI and HIT fields have evolved to date and also how they could evolve in the future. Our findings identify which research streams are central (versus peripheral) and which are cohesive (as opposed to disparate). Suggestions for vibrant areas of future research emerge from our analysis. The second part of the dissertation focuses on comprehensively understanding the effect of social support exchange in online health communities on individual members’ health promotion outcomes. This study examines the effectiveness of online consumer-to-consumer social support exchange on health promotion outcomes via analyses of big health data. Based on previous research, we propose a conceptual framework which integrates social capital theory and social support theory in the context of online health communities and test it through a quantitative field study and multiple analyses of a big online health community dataset. Specifically, natural language processing and machine learning techniques are utilized to automate content analysis of digital trace data. This research not only extends current theories of social support exchange in online health communities, but also sheds light on the design and management of such communities.

Radio frequency identification (RFID) technology is a useful technology that has myriad applications in technology, retail, manufacturing, and healthcare settings. Not dependent upon line-of-sight, RFID can scan devices in their proximity and report the information to connected (wired or other wireless) information systems. Once touted as the panacea for home healthcare, RFID devices can add benefit to patients in remote settings. RFID devices have been used to optimize systems in areas such as manufacturing and healthcare to expose inefficiencies in a system or process. Unlike manufacturing, however, RFID in healthcare settings presents security and privacy concerns to the people being tracked by the devices – particularly healthcare workers including nurses and doctors. This research presented a theoretical model that assessed the effect of five independent variables, namely, cognitive factors, of privacy concerns regarding surveillance and RFID devices and trust in the electronic medium, subjective norm, existence of security policy, and persistence of data on a dependent variable - intention to use RFID. The theoretical model presented in this research is based on the technology acceptance model and the extended theory of planned behavior. The research showed significant relationships between the cognitive factors of privacy concerns regarding surveillance and RFID devices, and trust and the electronic medium and perception of external control on intention to use. The theoretical model used in this research can be refined to better understand intention to use RFID in hospital environments.

This research examined if there is a relationship between engagement in quality improvement (QI) and health information technology (HIT) for local health departments (LHDs) controlling for workforce, finance, population, and governance structure. This was a cross-sectional study that analyzed data obtained from the Core questions and Module 1 in the NACCHO 2010 Profile of LHDs. Descriptive statistics, bivariate analyses, and logistic regression analyses were conducted. Findings suggest that LHD engagement in QI has a relationship with utilization of HIT including electronic health records, practice management systems, and electronic syndromic surveillance systems. This study provides baseline information about the HIT use of LHDs. LHDs and their system partners (hospitals, federally qualified health centers, and primary care providers) that utilize HIT as part of their QI decision making may have an easier time of using data to support evidence-based decision making and implementing the provisions of the Patient Protection and Affordable Care Act of 2010 in order to achieve population health for all.

The intention of implementing technology into healthcare practices is to reduce opportunity for errors in the delivery of providing health care. However, errors still occur, and many times are preventable. Configurations of health information technology systems should match clinical workflows to promote usage as intended. The purpose of this quality improvement project was to evaluate the impact of revised system configurations and use of a blood product transfusion system for the administration of blood products after one year of implementation. The method of heuristic evaluation is a usability engineering method for finding problems in a user interface design with the input of a small workgroup of subject matter experts. The project site had experienced reported incidents of blood product administration error as well as problems with systems communication since the implementation of the blood transfusion system. There were 31 nurse clinical educator staff users of the system who completed a survey evaluation of their perceptions of the blood transfusion system before and after configuration changes. The findings revealed that the mean quality and productivity score after the system configuration occurred was significantly higher than the mean score prior to the system configuration change, t (30) = -7.93, p < .001. The correlation between the one survey was also statistically significant, r = .46, p = .009. This project supports positive social change by reducing the potential for error for system users in the process of the blood administration process through heuristic evaluation through the implementation of changes to the technological system.

Racial/ethnic disparities in cancer care are well documented in the research literature; however, less is known about the extent and potential source of cancer care disparities in the Veterans Health Administration (VA). In my first paper, I use logistic regression and hospital fixed effects models to examine racial disparities in 20 cancer-related quality measures and the extent to which racial differences in site of care explain VA cancer care disparities. I found evidence of racial disparities in 7 out of 20 cancer-related quality measures. In general, these disparities were primarily driven by racial differences in care for black and white patients within the same VA hospital, rather than racial differences in site of care.

This thesis research contributes to national health Information Management/Information Technology (IM/IT) planning and therefore strategy development and implementation research, as well as to health information science. An examination into the national health IM/IT plans of several countries provides knowledge into identifying the typical IM/IT priorities that selected countries are focusing upon for healthcare improvement. Second, a systematic literature review of the current challenges, barriers and/or issues (referred to as ‘challenges’ hereafter) facing IM/IT priority implementation in healthcare settings provides insight on where nations should perhaps be focusing their attention, in order to enable more successful healthcare IM/IT implementations. Lastly, a study on national health IM/IT priorities contributes to the body of evidence that national level IM/IT direction is necessary for better patient care and health system reform across the world. In this investigation, the national health IM/IT priorities, which are reflected in the national health IM/IT strategic plans of five countries were assessed. To this end, the study: 1) Developed a set of measures to select four countries to study in addition to Canada; 2) Described the national health IM/IT priorities of Canada and four other countries; 3) Performed a systematic literature review of the challenges to overcome for successful implementation of IM/IT into healthcare settings; 4) Developed and administered a questionnaire where participants were asked to give their opinions on the progress their country has achieved in dealing with such challenges; and 5) Performed an analysis of the questionnaire results with respect to the countries’ national health IM/IT priorities. The systematic literature review uncovered a large number of challenges that the health informatics and healthcare community face when attempting to implement IM/IT into healthcare settings. iii The priority comparison highlighted that there is no right or wrong answer for what countries should focus their national health IM/IT energies upon. The findings indicate that nations focus their resources (time, money, personnel etc.) on the priorities they feel they should, whether those stem from needs analyses or politics. However, by learning about what other nations are prioritizing, a country can use that knowledge to help focus their own national health IM/IT priorities. The questionnaire results drew attention to the most frequently encountered challenges the five countries face in moving their national health IM/IT agendas forward. The feedback from the respondents provided individual reflections on how IM/IT implementations are actually progressing in their country, where problems are being encountered, including the nature of those problems, and in some cases, respondents offered insight on how to better deal with the challenges they face. The findings indicate that nations encounter similar problems in implementing IM/IT into healthcare settings. Currently, the world is facing many of the same healthcare system issues: shortages of healthcare processionals, long surgical and diagnostic imaging waitlists, ‘skyrocketing’ pharmaceutical drug pricing, healthcare funding practices, and challenges with implementing healthcare IM/IT priorities to name a few. If countries are facing similar health system problems, then it would be logical to assume that solutions to deal with such problems would be similar across nations. Thus, it is recommended that international fora and conferences be held to further discuss the types of health system IM/IT priorities that countries are implementing at a nation scale, the kinds of challenges they face and the solutions or conclusions that they have formulated in response to these challenges.