The Role of Data Science in Electronic Health Records: How Medical Decision Making can be improved based on a Comprehensive Electronic Medical Record?

Abstract

Despite the integration of modern technologies in medical applications, a significant gap remains in achieving high-level interaction between medical staff, physicians, and the systems they utilize. This gap often results in inefficiencies, user frustration, medical errors, and, in some cases, compromised patient safety, highlighting the critical need for improved system design. To address this issue, this thesis examines the impact of user interaction with these systems in medical settings, with a focus on the crucial role of Human-Computer Interaction (HCI) elements. A comprehensive systematic literature review (SLR) was conducted to identify and categorize HCI elements applicable within Clinical Decision Support System (CDSS) environments, emphasizing the necessity for Electronic Medical Records (EMRs) to be designed with these elements in mind, as they serve as the primary data source for CDSS. The current thesis extracted and categorized various HCI evaluation methods from existing studies based on their technical characteristics, providing a structured guideline for future investigations. Furthermore, the thesis details the impact of each HCI element on CDSS functionality, distinguishing between positive contributions and negative factors (termed "HCI barriers") that hinder effective interaction. Solutions to these barriers are also discussed in a dedicated chapter. Fundamentally, this thesis introduces a pivotal bridge between HCI principles and the critical domains of medical data management and quality. This foundational work has already led to the publication of three peer-reviewed scientific papers in prestigious journals, demonstrating its significant contribution to the field. Moreover, the benefits of integrating these HCI elements into other interconnected medical platforms, such as Personal Health Records (PHRs), were articulated. A novel cyclical EMR model is proposed that restructures patient data into distinct treatment cycles, thereby aligning digital records with the iterative nature of clinical workflows. This model enhances several critical HCI elements (including interface clarity, individuality, explainability, and user satisfaction) while improving data analysis and decision support accuracy. Empirical evaluations based on the proposed model reveal that structured data categorization and cyclebased data entry enhance the transparency and explainability of CDSS outputs, contributing to improved system usability and interpretability. Ultimately, this thesis presents a scientific framework that bridges the gap between HCI and medical data management, offering both theoretical insights and practical contributions to medical informatics. The significance of these contributions is further demonstrated by the publication of four peer-reviewed papers in prestigious journals, establishing a robust foundation for advancing CDSS development and user-centered system design in future research.