Kirstine Rosenbeck Gøeg

Content analysis of physical examination templates in electronic health records using SNOMED CT

OBJECTIVES Most electronic health record (EHR) systems are built on proprietary information models and terminology, which makes achieving semantic interoperability a challenge. Solving interoperability problems requires well-defined standards. In contrast, the need to support clinical work practice requires a local customization of EHR systems. Consequently, contrasting goals may be evident in EHR template design because customization means that local EHR organizations can define their own templates, whereas standardization implies consensus at some level. To explore the complexity of balancing these two goals, this study analyzes the differences and similarities between templates in use today. METHODS A similarity analysis was developed on the basis of SNOMED CT. The analysis was performed on four physical examination templates from Denmark and Sweden. The semantic relationships in SNOMED CT were used to quantify similarities and differences. Moreover, the analysis used these identified similarities to investigate the common content of a physical examination template. RESULTS The analysis showed that there were both similarities and differences in physical examination templates, and the size of the templates varied from 18 to 49 fields. In the SNOMED CT analysis, exact matches and terminology similarities were represented in all template pairs. The number of exact matches ranged from 7 to 24. Moreover, the number of unrelated fields differed a lot from 1/18 to 22/35. Cross-country comparisons tended to have more unrelated content than within-country comparisons. On the basis of identified similarities, it was possible to define the common content of a physical examination. Nevertheless, a complete view on the physical examination required the inclusion of both exact matches and terminology similarities. CONCLUSIONS This study revealed that a core set of items representing the physical examination templates can be generated when the analysis takes into account not only exact matches but also terminology similarities. This core set of items could be a starting point for standardization and semantic interoperability. However, both unmatched terms and terminology matched terms pose a challenge for standardization. Future work will include using local templates as a point of departure in standardization to see if local requirements can be maintained in a standardized framework.

Clustering clinical models from local electronic health records based on semantic similarity

BACKGROUND Clinical models in electronic health records are typically expressed as templates which support the multiple clinical workflows in which the system is used. The templates are often designed using local rather than standard information models and terminology, which hinders semantic interoperability. Semantic challenges can be solved by harmonizing and standardizing clinical models. However, methods supporting harmonization based on existing clinical models are lacking. One approach is to explore semantic similarity estimation as a basis of an analytical framework. Therefore, the aim of this study is to develop and apply methods for intrinsic similarity-estimation based analysis that can compare and give an overview of multiple clinical models. METHOD For a similarity estimate to be intrinsic it should be based on an established ontology, for which SNOMED CT was chosen. In this study, Lin similarity estimates and Sokal and Sneath similarity estimates were used together with two aggregation techniques (average and best-match-average respectively) resulting in a total of four methods. The similarity estimations are used to hierarchically cluster templates. The test material consists of templates from Danish and Swedish EHR systems. The test material was used to evaluate how the four different methods perform. RESULT AND DISCUSSION The best-match-average aggregation technique performed better in terms of clustering similar templates than the average aggregation technique. No difference could be seen in terms of the choice of similarity estimate in this study, but the finding may be different for other datasets. The dendrograms resulting from the hierarchical clustering gave an overview of the templates and a basis of further analysis. CONCLUSION Hierarchical clustering of templates based on SNOMED CT and semantic similarity estimation with best-match-average aggregation technique can be used for comparison and summarization of multiple templates. Consequently, it can provide a valuable tool for harmonization and standardization of clinical models.

Methods and applications for visualization of SNOMED CT concept sets

Inconsistent use of SNOMED CT concepts may reduce comparability of information in health information systems. Terminology implementation should be approached by common strategies for navigating and selecting proper concepts. This study aims to explore ways of illustrating common pathways and ancestors of particular sets of concepts, to support consistent use of SNOMED CT and also assess potential applications for such visualizations. The open source prototype presented is an interactive web-based re-implementation of the terminology visualization tool TermViz that provides an overview of concepts and their hierarchical relations. It provides terminological features such as interactively rearranging graphs, fetching more concept nodes, highlighting least common parents and shared pathways in merged graphs etc. Four teams of three to four people used the prototype to complete a terminology mapping task and then, in focus group interviews, discussed the user experience and potential future tool usage. Potential purposes discussed included SNOMED CT search and training, consistent selection of concepts and content management. The evaluation indicated that the tool may be useful in many contexts especially if integrated with existing systems, and that the graph layout needs further tuning and development.

A future-proof architecture for telemedicine using loose-coupled modules and HL7 FHIR

BACKGROUND AND OBJECTIVES Most telemedicine solutions are proprietary and disease specific which cause a heterogeneous and silo-oriented system landscape with limited interoperability. Solving the interoperability problem would require a strong focus on data integration and standardization in telemedicine infrastructures. Our objective was to suggest a future-proof architecture, that consisted of small loose-coupled modules to allow flexible integration with new and existing services, and the use of international standards to allow high re-usability of modules, and interoperability in the health IT landscape. METHODS We identified core features of our future-proof architecture as the following (1) To provide extended functionality the system should be designed as a core with modules. Database handling and implementation of security protocols are modules, to improve flexibility compared to other frameworks. (2) To ensure loosely coupled modules the system should implement an inversion of control mechanism. (3) A focus on ease of implementation requires the system should use HL7 FHIR (Fast Interoperable Health Resources) as the primary standard because it is based on web-technologies. RESULTS We evaluated the feasibility of our architecture by developing an open source implementation of the system called ORDS. ORDS is written in TypeScript, and makes use of the Express Framework and HL7 FHIR DSTU2. The code is distributed on GitHub. All modules have been tested unit wise, but end-to-end testing awaits our first clinical example implementations. CONCLUSIONS Our study showed that highly adaptable and yet interoperable core frameworks for telemedicine can be designed and implemented. Future work includes implementation of a clinical use case and evaluation.

Re-use of SNOMED CT subset in development of the Danish national standard for home care nursing problems

SNOMED CT was chosen as reference terminology for standardisation of homecare nursing documentation to make reporting comparable across the 98 Danish municipalities. The method outlined in this paper for developing a Danish national homecare nursing SNOMED CT subsets is a pragmatic approach to build new SNOMED CT subsets drawing on existing and available SNOMED CT subsets. Combining this approach with awareness of hierarchical coherency in SNOMED CT subsets makes effective retrieval of data possible.

Visualizing sets of SNOMED CT concepts to support consistent terminology implementation and reuse of clinical data

Inconsistent use of concepts is an obstacle when implementing SNOMED CT to improve comparability of information. Terminology implementation should be approached by common strategies for navigating and selecting proper concepts. This study aims to explore ways of illustrating common pathways and ancestors of particular sets of concepts, to support consistent use of SNOMED CT in EHR-system implementation processes. The prototype presented is an interactive web-based re-implementation of the terminology visualization tool TermViz. The open source prototype contains terminological features that are of relevance when exploring and comparing sets of concepts in SNOMED CT. Future work should focus on evaluating the developed prototype in order to assess its applicability in EHR-system-implementation contexts.

Feasibility of representing a Danish microbiology model Using FHIR

Achieving interoperability in health is a challenge and requires standardization. The newly developed HL7 standard: Fast Healthcare Interoperability Resources (FHIR) promises both flexibility and interoperability. This study investigates the feasibility of expressing a Danish microbiology message model content in FHIR to explore whether complex in-use legacy models can be migrated and what challenges this may pose. The Danish microbiology message model (the DMM) is used as a case to illustrate challenges and opportunities accosted with applying the FHIR standard. Mapping of content from DMM to FHIR was done as close as possible to the DMM to minimize migration costs except when the structure of the content did not fit into FHIR. From the DMM a total of 183 elements were mapped to FHIR. 75 (40.9%) elements were modeled as existing FHIR elements and 96 (52.5%) elements were modeled as extensions and 12 (6.6%) elements were deemed unnecessary because of build-in FHIR characteristics. In this study, it was possible to represent the content of a Danish message model using HL7 FHIR.

Developing an acute-physical-examination template for a Tegional EHR system aimed at improving inexperienced physician's documentation.

It is well-established that to increase acceptance of electronic clinical documentation tools, such as electronic health record (EHR) systems, it is important to have a strong relationship between those who document the clinical encounters and those who reaps the benefit of digitalized and more structured documentation. [1] Therefore, templates for EHR systems benefit from being closely related to clinical practice with a strong focus on primarily solving clinical problems. Clinical use as a driver for structured documentation has been the focus of the acute-physical-examination template (APET) development in the North Denmark Region. The template was developed through a participatory design where precision and clarity of documentation was prioritized as well as fast registration. The resulting template has approximately 700 easy accessible input possibilities and will be evaluated in clinical practice in the first quarter of 2013.

SNOMED CT as reference terminology in the Danish national home care documentation standard

In Danish home care, multiple professions deliver services to citizens. FSIII is a national home care documentation standard, where one of the goals is to share documentation to improve coordination between these professional groups and avoid double documentation. The aim of this study was to develop a SNOMED-CT based navigation hierarchy to ensure that professions could preserve their documentation practice, to help avoid double documentation, and to ensure that the technical implementation did not require sophisticated semantic tools. The method involved mapping of non-SNOMED-CT content to SNOMED CT, visualization of merged graphs, identification of reference concepts, relating reference concepts to the documentation models of each profession, and representation of the navigation hierarchy in a reference set. The navigation hierarchy ensures that citizen conditions appear in a relevant context, regardless of which profession entered the data. Our approach paves the way for incremental standardization projects, where an implementation artefact, such as the navigation hierarchy, highlights the semantic features of SNOMED CT that can be used to reach specific business goals; in this case, sharing data across professional groups.

The costs and benefits of SNOMED CT implementation: an economic assessment model

Health—exploring complexity: an interdisciplinary systems approach HEC2016 28 August–2 September 2016, Munich, Germany Eva Grill • Martin Müller • Ulrich Mansmann Springer Science+Business Media Dordrecht 2016 Health is a complex process with potentially interacting components from the molecular to the societal and environmental level. Adequate research designs and data analysis methods are needed to improve our understanding of this complexity, to ultimately derive high quality evidence to inform patients, health professionals, and health policy decision makers. Also, effective patient-centred health care has to address the complexity of health, functioning, and disability, not only by implementing interventions, but also by using information technology that represents the complexity of health care to inform all actors. Given this background, we developed the concept of our conference HEC2016 as an interdisciplinary European event in beautiful Bavaria, in the city of München. Quite ironically this is the place, where William of Ockham, whose ideas of parsimony are the very opposite of complexity, accused of heresy, spent 17 years under the protection of the Bavarian King Ludwig IV. Furthermore, our local public health hero Max von Pettenkofer (1818–1901) contributed a lot to the basic systemic understanding of health, especially the relevance of a healthy environment. Under the joint theme of health as a complex system we joined the activities of five scientific disciplines: Medical Informatics, Medical Biometry, Bioinformatics, Epidemiology and Health Data Management. The mission behind this interdisciplinary effort was to serve as an important scientific forum for the exchange of new ideas and applications to strengthen health sciences on a national and international level. The analysis of health as a complex system opens needed perspectives on a challenging reality: filtering current hypotheses, resolving controversies, and tailoring interventions to the need of the individual within a health system environment. The conference encouraged the dialogue of the disciplines in order to advance our understanding of health and to decrease burden of disease. HEC2016 brought together the annual conferences of the German Association for Medical Informatics, Biometry and Epidemiology (GMDS), the German Society for Epidemiology (DGEpi), the International Epidemiological AssociationEuropean Region (IEA-EEF) and the European Federation for Medical Informatics Association (EFMI, MIE 2016). HEC2016 took place in München, Germany, in the main building of the Ludwig-Maximilians-Universität (LMU) from 28 August to 2 September 2016 under the auspices of the Institute for Medical Information Processing, Biometry and Epidemiology of LMU. The conference received 832 contributions for oral and poster presentation (Table 1). Fourteen percent of them were from outside Europe with the largest group of 10 % from Asia (Table 2). Scientific program committees and reviewers selected 408 submissions as oral contributions and 303 for poster presentations. The program was surrounded by twelve tutorials held by international renowned scientists and covered a broad spectrum from innovative biostatistical and epidemiological methods to tutorials in application of innovative software, scientific writing and data protection issues. Over 50 panel discussions and workshops allowed in-depth exchange of ideas on specific topics and underscored the interactive nature of HEC2016. A special focus of HEC2016 was on the promotion of young scientists from all disciplines whose participation was supported by numerous travel grants. We would like to express our deepest gratitude to all the colleagues who supported us as speakers, committee members and reviewers, lent us a hand before, during and after the conference, gave critical but friendly comments at all stages of the preparations, supported us by providing coffee, audience or Butterbrezen, and specifically to those who submitted contributions to the conference and attended the conference and its many tutorials, lectures and sessions. We extend our gratitude to the Deutsche Forschungsgemeinschaft for generous financial support (grant no. GR 3608/4-1). Last not least we would like to thank our families who allowed us to spend most of our weekends with organizing this conference, to William of Ockham for lending us his razor (from time to time) and to Max von Pettenkofer for guidance. Eva Grill Martin Müller Ulrich Mansmann for the local organizing committee (Tables 1, 2) 123 Eur J Epidemiol (2016) 31:S1–S239 DOI 10.1007/s10654-016-0183-1Original citation: Lim Choi Keung, Sarah Niukyun, Murphy, Paddie, Khan, Mohammed Omar, Perkins, Gavin D., Smith, Christopher and Arvanitis, Theodoros N. (2016) Perspectives of UK community first responders on a national public access defibrillator database. In: HEC 2016 (Health Exploring Complexity: An Interdisciplinary Systems Approach) Conference, Munich, Germany, 28 Aug 2 Sep 2016. Published in: European Journal of Epidemiology, 31 (Supplement 1). S117 (796).