Deep learning models are increasingly used to analyze health records to model disease progression. Two characteristics of health records present challenges to developers of deep learning-based medical systems. First, the veracity of the estimation of missing health data must be evaluated to optimize the performance of deep learning models. Second, the currently most successful deep learning diagnostic models, called transformers, lack a mechanism to analyze the temporal characteristics of health records. In this thesis, these two challenges are investigated using a real-world medical dataset of longitudinal health records from 340,143 patients over ten years called MIIDD: McMaster Imaging Information and Diagnostic Dataset. To address missing data, the performance of imputation models (mean, regression, and deep learning) were evaluated on a real-world medical dataset. Next, techniques from adversarial machine learning were used to demonstrate how imputation can have a cascading negative impact on a deep learning model. Then, the strengths and limitations of evaluation metrics from the statistical literature (qualitative, predictive accuracy, and statistical distance) to evaluate deep learning-based imputation models were investigated. This research can serve as a reference to researchers evaluating the impact of imputation on their deep learning models. To analyze the temporal characteristics of health records, a new model was developed and evaluated called DTTHRE: Decoder Transformer for Temporally-Embedded Health Records Encoding. DTTHRE predicts patients' primary diagnoses by analyzing their medical histories, including the elapsed time between visits. The proposed model successfully predicted patients' primary diagnosis in their final visit with improved predictive performance (78.54 +/- 0.22%) compared to existing models in the literature. DTTHRE also increased the training examples available from limited medical datasets by predicting the primary diagnosis for each visit (79.53 +/- 0.25%) with no additional training time. This research contributes towards the goal of disease predictive modeling for clinical decision support. / Dissertation / Doctor of Philosophy (PhD) / In this thesis, two challenges using deep learning models to analyze health records are investigated using a real-world medical dataset. First, an important step in analyzing health records is to estimate missing data. We investigated how imputation can have a cascading negative impact on a deep learning model's performance. A comparative analysis was then conducted to investigate the strengths and limitations of evaluation metrics from the statistical literature to assess deep learning-based imputation models. Second, the most successful deep learning diagnostic models to date, called transformers, lack a mechanism to analyze the temporal characteristics of health records. To address this gap, we developed a new temporally-embedded transformer to analyze patients' medical histories, including the elapsed time between visits, to predict their primary diagnoses. The proposed model successfully predicted patients' primary diagnosis in their final visit with improved predictive performance (78.54 +/- 0.22%) compared to existing models in the literature.

No description available.

No description available.

No description available.

It has been well established, in the literature, the association between low health literacy rates and poor health outcomes. With the increase of technology dependence, more people are using the internet to look up health information. Research has shown that shared decision making between providers and patients can improve patients' health outcomes. This research aims to examine whether electronic health (eHealth) literacy impacts patient-provider relationships. This research will also examine how geography specifically state residency impacts eHealth literacy rates. Data collected from a national sampling of online health and medical information users who participated in the Study of Health and Medical Information in Cyberspace (N=710) is used to construct structural equation models from SPSS AMOS v. 20.0. After path analysis, the results shown that white males with higher education were more likely to have higher eHealth literacy rates and that eHealth literacy rates are associated with better self-care, higher patient satisfaction and increased trust in provider. Also, state residency does not have an impact on eHealth literacy rates. eHealth literacy will be significant in patient-provider relationships. Program development should be established on focusing on eHealth literacy across the lifespan. Also, it will be important to review federal policy on technology disbursements in order to achieve national goals on eHealth literacy rates.

The aim of this project was to investigate what information is critical to convey to nurses when performing digital triage. In addition, the project aimed to investigate how such information could be visualized. This was done through a combined user research and machine learning approach, which enabled for a more nuanced and thorough investigation compared to only making use of one of the two fields. There is sparse research investigating how digital triaging can be improved and made more efficient. Therefore, this study has contributed with new and relevant insights. Three machine learning algorithms were implemented to predict the right level of care for a patient. Out of these three, the random forest classifier proved to have the best performance with an accuracy of 69.46%, also having the shortest execution time. Evaluating the random forest classifier, the most important features were stated to be the duration and progress of the symptoms, allergies to medicine, chronic diseases and the patient's own estimation of his/her health. These factors could all be confirmed by the user research approach, indicating that the results from the approaches were aligned. The results from the user research approach also showed that the patients' own description of their symptoms was of great importance. These findings served as a basis for a number of visualization decisions, aiming to make the triage process as accurate and efficient as possible.

Em 2005, a Agência Nacional de Saúde Suplementar (ANS) estabelece o padrão TISS (Troca de Informação na Saúde Suplementar), intercâmbio eletrônico obrigatório entre as operadoras de planos de saúde (cerca de 1500 registradas na ANS) e prestadores de serviços (cerca de 200 mil) sobre os eventos de assistência prestados aos beneficiários. O padrão TISS foi desenvolvido seguindo a estrutura do Comitê ISO/TC215 de padrões para informática em saúde e se divide em quatro partes: conteúdo e estrutura, que compreende a estrutura das guias em papel; representação de conceitos em saúde, que se refere às tabelas de domínio e vocabulários em saúde; comunicação, que contempla as mensagens eletrônicas; e segurança e privacidade, seguindo recomendação do Conselho Federal de Medicina (CFM). Para aprimorar sua metodologia de evolução, essa presente tese analisou o grau de interoperabilidade do padrão TISS segundo a norma ISO 20514 (ISO 20514, 2005) e a luz do modelo dual da Fundação openEHR, que propõe padrões abertos para arquitetura e estrutura do Registro Eletrônico de Saúde (RES). O modelo dual da Fundação openEHR é composto, no primeiro nível, por um modelo de referência genérico e, no segundo, por um modelo de arquétipos, onde se detalham os conceitos e atributos. Dois estudos foram realizados: o primeiro se refere a um conjunto de arquétipos demográficos elaborados como proposta de representação da informação demográfica em saúde, baseado no modelo de referência da Fundação openEHR. O segundo estudo propõe um modelo de referência genérico, como aprimoramento das especificações da Fundação openEHR, para representar o conceito de submissão de autorização e contas na saúde, assim como um conjunto de arquétipos. Por fim, uma nova arquitetura para construção do padrão TISS é proposta, tendo como base o modelo dual da Fundação openEHR e como horizonte a evolução para o RES centrado no paciente / In 2005 the Brazilian National Healthcare Agency (Agência Nacional de Saúde Suplementar (ANS), in Portuguese) published the TISS standard, a mandatory electronic exchange claims between the health insurance (approximately 1500 registered at ANS) and healthcare providers (approximately 200 000) about the healthcare events provided to the beneficiaries. The TISS standard was developed following the structure of the ISO/TC215 Committee for Health Informatics and is divided into four parts: data structure that encompasses the forms; semantic content that refers to the vocabularies and terminologies; data interchange that includes the electronic messages; and security issues in healthcare information following a recommendation of the Federal Council of Medicine (Conselho Federal de Medicina (CFM), in Portuguese). To improve the TISS standards evolution, this study analyses the levels of interoperability in accordance with ISO 20514 (ISO 20514, 2005) and the dual model of the openEHR Foundation, which proposes open standards for Electronic Health Records (EHR) architecture and healthcare information. The dual model of the openEHR Foundation specifications is composed, on the first level, by a generic reference model and on the second by an archetype model that details the concepts and attributes. Two studies were conducted in this thesis: the first one refers to a set of demographic archetypes developed as a proposal representation of the demographic information, based on the reference model of the openEHR Foundation. The second study proposes a generic reference model, as an improvement of the openEHR specifications, to represent the concept of submission of claims, as well as a set of archetypes. Finally, a new architecture for building the TISS standard is proposed based on the dual model of the openEHR Foundation and envisioning a patient-centered EHR

<p>Dagens elektroniska patientjournalsystem har begränsade möjligheter att på likartat sätt strukturera och lagra patientinformation. Det är en anledning till att det är problem med att utbyta patientjournaldata mellan olika system. Detta försvårar bland annat forskning och tillgänglighet till patientinformation. Brist på tillgänglighet minskar i sin tur möjligheten att ge en god vård oberoende av var patienten befinner sig.</p><p>Inom projektet openEHR har en idé med så kallade arketyper tagits fram som ett enhetligt sätt att strukturera utbytbar patientjournaldata för att möta framtida krav på patientjournaler och patientjournalsystem. Arketyper är formella modeller av kliniska informationsentiteter, exempelvis blodtryck. De byggs upp av restriktioner, struktur och termer med eventuella bindningar till medicinska terminologisystem. Dessutom kopplas medicinsk kunskap i arketyperna fri från journalsystemen.</p><p>Syftet med examensarbetet har varit att utveckla ett verktyg, en så kallad arketypeditor, som kan användas för att skapa och redigera arketyper. Utöver detta skulle möjligheterna undersökas att i verktyget implementera en koppling till medicinska terminologisystem. Utvecklingen har skett i en iterativ process med fokus på användbarhet och stabilitet. Det har även ingått att ta reda på syftet med en arketypeditor.</p><p>Resultatet är ett plattformsoberoende och stabilt verktyg som är utvecklat enligt användbarhetsprinciper med koppling till terminologisystemet Unified Medical Language System (UMLS). En arketypeditors syfte i ett bredare perspektiv är att lösa brister i dagens medicinska informationssystem som tas upp i denna rapport. Trots att openEHR-projektet är nytt finns det många tekniskt gångbara idéer, men det finns även problem som beror på för lite praktisk testning och tillämpning.</p> / <p>Present-day electronic health record systems have limited possibilities to structure and store patient information in a similarly manner. This causes problems with exchanging patient record data between different systems and it gives rise to problems with, among other things, research and patient information availability. Lack of availability will in turn decrease the possibility of giving good care irrespective of where the patient is located.</p><p>Within the openEHR project an idea with so called archetypes has been introduced as a uniform way to structure exchangeable patient record data in order to meet future requirements on electronic health records and systems. Archetypes are formal models of clinical information entities, for example blood pressure. They are constructed from constraints, structure and terms which may have bindings to medical terminology systems. Furthermore, medical knowledge in the archetypes is separated from the patient record systems.</p><p>The purpose of the thesis has been to develop a tool, a so called archetype editor, that can be used to create and edit archetypes. In addition, the possibilities of implementing a connection to medical terminology systems should be explored. The development has followed an iterative process with focus on stability and usability. Another task has also been to find out the purpose with an archetype editor.</p><p>The result is a platform-independent and stable tool, developed according to usability principles with a connection to the terminology system Unified Medical Language System (UMLS). An archetype editor’s purpose in a wider perspective is to solve shortcomings in medical information systems of today, which are brought up in this thesis. Although the openEHR project is new, there are many technically applicable ideas but also problems because of insufficient practical testing and application.</p>

Several factors are driving the transition from paper-based health records to electronic health record systems. In the United States, the adoption rate of electronic health record systems significantly increased after "Meaningful Use" incentive program was started in 2009. While increased use of electronic health record systems could improve the efficiency and quality of healthcare services, it can also lead to a number of security and privacy issues, such as identity theft and healthcare fraud. Such incidents could have negative impact on trustworthiness of electronic health record technology itself and thereby could limit its benefits. In this dissertation, we tackle three challenges that we believe are important to improve the security and privacy in electronic health record systems. Our approach is based on an analysis of real-world incidents, namely theft and misuse of patient identity, unauthorized usage and update of electronic health records, and threats from insiders in healthcare organizations. Our contributions include design and development of a user-centric monitoring agent system that works on behalf of a patient (i.e., an end user) and securely monitors usage of the patient's identity credentials as well as access to her electronic health records. Such a monitoring agent can enhance patient's awareness and control and improve accountability for health records even in a distributed, multi-domain environment, which is typical in an e-healthcare setting. This will reduce the risk and loss caused by misuse of stolen data. In addition to the solution from a patient's perspective, we also propose a secure system architecture that can be used in healthcare organizations to enable robust auditing and management over client devices. This helps us further enhance patients' confidence in secure use of their health data.

Dagens elektroniska patientjournalsystem har begränsade möjligheter att på likartat sätt strukturera och lagra patientinformation. Det är en anledning till att det är problem med att utbyta patientjournaldata mellan olika system. Detta försvårar bland annat forskning och tillgänglighet till patientinformation. Brist på tillgänglighet minskar i sin tur möjligheten att ge en god vård oberoende av var patienten befinner sig. Inom projektet openEHR har en idé med så kallade arketyper tagits fram som ett enhetligt sätt att strukturera utbytbar patientjournaldata för att möta framtida krav på patientjournaler och patientjournalsystem. Arketyper är formella modeller av kliniska informationsentiteter, exempelvis blodtryck. De byggs upp av restriktioner, struktur och termer med eventuella bindningar till medicinska terminologisystem. Dessutom kopplas medicinsk kunskap i arketyperna fri från journalsystemen. Syftet med examensarbetet har varit att utveckla ett verktyg, en så kallad arketypeditor, som kan användas för att skapa och redigera arketyper. Utöver detta skulle möjligheterna undersökas att i verktyget implementera en koppling till medicinska terminologisystem. Utvecklingen har skett i en iterativ process med fokus på användbarhet och stabilitet. Det har även ingått att ta reda på syftet med en arketypeditor. Resultatet är ett plattformsoberoende och stabilt verktyg som är utvecklat enligt användbarhetsprinciper med koppling till terminologisystemet Unified Medical Language System (UMLS). En arketypeditors syfte i ett bredare perspektiv är att lösa brister i dagens medicinska informationssystem som tas upp i denna rapport. Trots att openEHR-projektet är nytt finns det många tekniskt gångbara idéer, men det finns även problem som beror på för lite praktisk testning och tillämpning. / Present-day electronic health record systems have limited possibilities to structure and store patient information in a similarly manner. This causes problems with exchanging patient record data between different systems and it gives rise to problems with, among other things, research and patient information availability. Lack of availability will in turn decrease the possibility of giving good care irrespective of where the patient is located. Within the openEHR project an idea with so called archetypes has been introduced as a uniform way to structure exchangeable patient record data in order to meet future requirements on electronic health records and systems. Archetypes are formal models of clinical information entities, for example blood pressure. They are constructed from constraints, structure and terms which may have bindings to medical terminology systems. Furthermore, medical knowledge in the archetypes is separated from the patient record systems. The purpose of the thesis has been to develop a tool, a so called archetype editor, that can be used to create and edit archetypes. In addition, the possibilities of implementing a connection to medical terminology systems should be explored. The development has followed an iterative process with focus on stability and usability. Another task has also been to find out the purpose with an archetype editor. The result is a platform-independent and stable tool, developed according to usability principles with a connection to the terminology system Unified Medical Language System (UMLS). An archetype editor’s purpose in a wider perspective is to solve shortcomings in medical information systems of today, which are brought up in this thesis. Although the openEHR project is new, there are many technically applicable ideas but also problems because of insufficient practical testing and application.