Programming techniques for improving rule readability for rule-based information extraction natural language processing pipelines of unstructured and semi-structured medical texts.

BACKGROUND: Extraction of medical terms and their corresponding values from semi-structured and unstructured texts of medical reports can be a time-consuming and error-prone process. Methods of natural language processing (NLP) can help define an extraction pipeline for accomplishing a structured format transformation strategy. OBJECTIVES: In this paper, we build an NLP pipeline to extract values of the classification of malignant tumors (TNM) from unstructured and semi-structured pathology reports and import them further to a structured data source for a clinical study. Our research interest is not focused on standard performance metrics like precision, recall, and F-measure on the test and validation data. We discuss how with the help of software programming techniques the readability of rule-based (RB) information extraction (IE) pipelines can be improved, and therefore minimize the time to correct or update the rules, and efficiently import them to another programming language. METHODS: The extract rules were manually programmed with training data of TNM classification and tested in two separate pipelines based on design specifications from domain experts and data curators. Firstly we implemented each rule directly in one line for each extraction item. Secondly, we reprogrammed them in a readable fashion through decomposition and intention-revealing names for the variable declaration. To measure the impact of both methods we measure the time for the fine-tuning and programming of the extractions through test data of semi-structured and unstructured texts. RESULTS: We analyze the benefits of improving through readability of the writing of rules, through parallel programming with regular expressions (REGEX), and the Apache Uima Ruta language (AURL). The time for correcting the readable rules in AURL and REGEX was significantly reduced. Complicated rules in REGEX are decomposed and intention-revealing declarations were reprogrammed in AURL in 5 min. CONCLUSION: We discuss the importance of factor readability and how can it be improved when programming RB text IE pipelines. Independent of the features of the programming language and the tools applied, a readable coding strategy can be proven beneficial for future maintenance and offer an interpretable solution for understanding the extraction and for transferring the rules to other domains and NLP pipelines.

Structured, Harmonized, and Interoperable Integration of Clinical Routine Data to Compute Heart Failure Risk Scores.

Risk prediction in patients with heart failure (HF) is essential to improve the tailoring of preventive, diagnostic, and therapeutic strategies for the individual patient, and effectively use health care resources. Risk scores derived from controlled clinical studies can be used to calculate the risk of mortality and HF hospitalizations. However, these scores are poorly implemented into routine care, predominantly because their calculation requires considerable efforts in practice and necessary data often are not available in an interoperable format. In this work, we demonstrate the feasibility of a multi-site solution to derive and calculate two exemplary HF scores from clinical routine data (MAGGIC score with six continuous and eight categorical variables; Barcelona Bio-HF score with five continuous and six categorical variables). Within HiGHmed, a German Medical Informatics Initiative consortium, we implemented an interoperable solution, collecting a harmonized HF-phenotypic core data set (CDS) within the openEHR framework. Our approach minimizes the need for manual data entry by automatically retrieving data from primary systems. We show, across five participating medical centers, that the implemented structures to execute dedicated data queries, followed by harmonized data processing and score calculation, work well in practice. In summary, we demonstrated the feasibility of clinical routine data usage across multiple partner sites to compute HF risk scores. This solution can be extended to a large spectrum of applications in clinical care.

openEHR-to-FHIR: Converting openEHR Compositions to Fast Healthcare Interoperability Resources (FHIR) for the German Corona Consensus Dataset (GECCO).

The German Corona Consensus (GECCO) established a uniform dataset in FHIR format for exchanging and sharing interoperable COVID-19 patient specific data between health information systems (HIS) for universities. For sharing the COVID-19 information with other locations that use openEHR, the data are to be converted in FHIR format. In this paper, we introduce our solution through a web-tool named "openEHR-to-FHIR" that converts compositions from an openEHR repository and stores in their respective GECCO FHIR profiles. The tool provides a REST web service for ad hoc conversion of openEHR compositions to FHIR profiles.

vivaGen - a survival data set generator for software testing.

An amendment to this paper has been published and can be accessed via the original article.

Patient Survival After Kidney Transplantation: Important Role of Graft-sustaining Factors as Determined by Predictive Modeling Using Random Survival Forest Analysis.

BACKGROUND: Identification of the relevant factors for death can improve patient's individual risk assessment and decision making. A well-documented patient cohort (n = 892) in a renal transplant program with protocol biopsies was used to establish multivariable models for risk assessment at 3 and 12 months posttransplantation by random survival forest analysis. METHODS: Patients transplanted between 2000 and 2007 were observed for up to 11 years. Loss to follow-up was negligible (n = 15). A total of 2251 protocol biopsies and 1214 biopsies for cause were performed. All rejections and clinical borderline rejections in protocol biopsies were treated. RESULTS: Ten-year patient survival was 78%, with inferior survival of patients with graft loss. Using all pre- and posttransplant variables until 3 and 12 months (n = 65), the obtained models showed good performance to predict death (concordance index: 0.77-0.78). Validation with a separate cohort of patients (n = 349) showed a concordance index of 0.76 and good discrimination of risks by the models, despite substantial differences in clinical variables. Random survival forest analysis produced robust models over a wide range of parameter settings. Besides well-established risk factors like age, cardiovascular disease, type 2 diabetes, and graft function, posttransplant urinary tract infection and rejection treatment were important factors. Urinary tract infection and rejection treatment were not specifically associated with death due to infection or malignancy but correlated strongly with inferior graft function and graft loss. CONCLUSIONS: The established models indicate the important areas that need special attention in the care of renal transplant patients, particularly modifiable factors like graft rejection and urinary tract infection.

Methods for a similarity measure for clinical attributes based on survival data analysis.

BACKGROUND: Case-based reasoning is a proven method that relies on learned cases from the past for decision support of a new case. The accuracy of such a system depends on the applied similarity measure, which quantifies the similarity between two cases. This work proposes a collection of methods for similarity measures especially for comparison of clinical cases based on survival data, as they are available for example from clinical trials. METHODS: Our approach is intended to be used in scenarios, where it is of interest to use longitudinal data, such as survival data, for a case-based reasoning approach. This might be especially important, where uncertainty about the ideal therapy decision exists. The collection of methods consists of definitions of the local similarity of nominal as well as numeric attributes, a calculation of attribute weights, a feature selection method and finally a global similarity measure. All of them use survival time (consisting of survival status and overall survival) as a reference of similarity. As a baseline, we calculate a survival function for each value of any given clinical attribute. RESULTS: We define the similarity between values of the same attribute by putting the estimated survival functions in relation to each other. Finally, we quantify the similarity by determining the area between corresponding curves of survival functions. The proposed global similarity measure is designed especially for cases from randomized clinical trials or other collections of clinical data with survival information. Overall survival can be considered as an eligible and alternative solution for similarity calculations. It is especially useful, when similarity measures that depend on the classic solution-describing attribute "applied therapy" are not applicable. This is often the case for data from clinical trials containing randomized arms. CONCLUSIONS: In silico evaluation scenarios showed that the mean accuracy of biomarker detection in k = 10 most similar cases is higher (0.909-0.998) than for competing similarity measures, such as Heterogeneous Euclidian-Overlap Metric (0.657-0.831) and Discretized Value Difference Metric (0.535-0.671). The weight calculation method showed a more than six times (6.59-6.95) higher weight for biomarker attributes over non-biomarker attributes. These results suggest that the similarity measure described here is suitable for applications based on survival data.

Implementing LOINC - Current Status and Ongoing Work at a Medical University.

The Logical Observation Identifiers, Names and Codes (LOINC) is a common terminology used for standardizing laboratory terms. Within the consortium of the HiGHmed project, LOINC is one of the central terminologies used for health data sharing across all university sites. Therefore, linking the LOINC codes to the site-specific tests and measures is one crucial step to reach this goal. In this work we report our ongoing efforts in implementing LOINC to our laboratory information system and research infrastructure, as well as our challenges and the lessons learned. 407 local terms could be mapped to 376 LOINC codes of which 209 are already available to routine laboratory data. In our experience, mapping of local terms to LOINC is a widely manual and time consuming process for reasons of language and expert knowledge of local laboratory procedures.

Clinical Information Model Based Data Quality Checks: Theory and Example.

INTRODUCTION: We describe principles of leveraging clinical information models (CIMs) for data quality (DQ) checks and present the exemplary application of these principles. METHODS: openEHR compliant CIMs are used to express DQ-checks as constraints. Test setting is the process of extracting, transforming and loading (ETL) assisted ventilation data from two patient data management systems (PDMS) of a pediatric intensive care unit into a local openEHR-based data repository. RESULTS: A generic component logs aggregated DQ-check results for ~28 million entries. DQ-issue types in the presented results are range-, format- and value set violations. DISCUSSION: CIMs are suitable means to define DQ-checks for range-, format-, value set and cardinality constraints. However, they cannot constitute a complete solution for standardized DQ-assessment.

A Report on Archetype Modelling in a Nationwide Data Infrastructure Project.

BACKGROUND: The nationwide data infrastructure project HiGHmed strives for achieving semantic interoperability through the use of openEHR archetypes. Therefore, a knowledge governance framework defining collaborative modelling processes has been established. For long-sustained success and the creation of high-quality archetypes, continuous monitoring is vital. OBJECTIVES: To present an update on archetype modelling and governance framework establishment in HiGHmed. METHODS: Qualitative and quantitative analyses of the progress in establishing modelling groups, roles and users, realizing modelling workflows, and modelling archetypes. RESULTS: Currently, 25 modellers and 17 domain experts are participating. 79 archetypes have been identified, from which 69 are pre-existing and internationally published; completion rates of review rounds are satisfying but improvable. CONCLUSIONS: The governance framework is valuable to make the activities manageable and to accelerate modelling. Combined with highly engaged data stewards and clinicians, a reasonable number of archetypes have already been developed.

Implementing LOINC: Current Status and Ongoing Work at the Hannover Medical School.

The Logical Observation Identifiers, Names and Codes (LOINC) is a common terminology used for standardizing laboratory terms. Within the HiGHmed consortium, LOINC is used as a central terminology for health data sharing across all university hospital sites. Therefore, linking the LOINC codes to the site-specific tests and measures is one crucial step to reach this goal. In this work we report our ongoing work in implementing LOINC to the laboratory information system, our challenges and lessons learned.

Implementing Systems Medicine: A Medical Informatics Perspective.

Systems medicine is a paradigm for translating in silico methods developed for modelling biological systems into the field of medicine. Such approaches rely on the integration of as many data sources as possible, both in the dimension of disease knowledge and patient data. This is a challenging task that can only be implemented in clinical routine with the help of suitable information technology from the field of Medical Informatics. For the research project "Clinically-applicable, omics-based assessment of survival, side effects, and targets in multiple myeloma" (CLIOMMICS) we developed a prototypical systems medicine application system. It is based on a three-level-architecture covering data representation, decision support, and user interface. The core decision support component is implemented as a case-based reasoning engine. However, the architecture follows a modular design that allows to replace individual components as needed.

On Teaching International Courses on Health Information Systems. Lessons Learned during 16 Years of Frank - van Swieten Lectures on Strategic Information Management in Health Information Systems.

BACKGROUND: Health information systems (HIS) are one of the most important areas for biomedical and health informatics. In order to professionally deal with HIS well-educated informaticians are needed. Because of these reasons, in 2001 an international course has been established: The Frank - van Swieten Lectures on Strategic Information Management of Health Information Systems. OBJECTIVES: Reporting about the Frank - van Swieten Lectures and about our students' feedback on this course during the last 16 years. Summarizing our lessons learned and making recommendations for such international courses on HIS. METHODS: The basic concept of the Frank - van Swieten lectures is to teach the theoretical background in local lectures, to organize practical exercises on modelling sub-information systems of the respective local HIS and finally to conduct Joint Three Days as an international meeting were the resulting models are introduced and compared. RESULTS: During the last 16 years, the Universities of Amsterdam, Braunschweig, Heidelberg/Heilbronn, Leipzig as well as UMIT were involved in running this course. Overall, 517 students from these universities participated. Our students' feedback was clearly positive. The Joint Three Days of the Frank - van Swieten Lectures, where at the end of the course all students can meet, turned out to be an important component of this course. Based on the last 16 years, we recommend common teaching materials, agreement on equivalent clinical areas for the exercises, support of group building of international student groups, motivation of using a collaboration platform, ensuring quality management of the course, addressing different levels of knowledge of the students, and ensuring sufficient funding for joint activities. CONCLUSIONS: Although associated with considerable additional efforts, we can clearly recommend establishing such international courses on HIS, such as the Frank - van Swieten Lectures.

The Use of Tools, Modelling Methods, Data Types, and Endpoints in Systems Medicine: A Survey on Projects of the German e:Med-Programme.

Systems medicine is the consequent continuation of research efforts on the road to an individualized medicine. Thereby, systems medicine tries to offer a holistic view on the patient by combining different data sources to highlight different perspectives on the patient's health. Our research question was to identify the main data types, modelling methods, analysis tools, and endpoints currently used and studied in systems medicine. Therefore, we conducted a survey on projects with a systems medicine background. Fifty participants completed this survey. The results of the survey were analyzed using histograms and cross tables, and finally compared to results of a former literature review with the same research focus. The data types reported in this survey were widely diversified. As expected, genomic and phenotype data were used most frequently. In contrast, environmental and behavioral data were rarely used in the projects. Overall, the cross tables of the data types in the survey and the literature review showed overlapping results.

[An urban EMS at the start of a cross-sectoral quality management system: prioritized implementation of the 2010 ERC recommendations and long-term survival after cardiac arrest]. / Ein städtischer Rettungsdienstbereich am Beginn eines sektorenübergreifenden Qualitätsmanagementsystems: Eine priorisierte Umsetzung der ERC Empfehlungen von 2010 und das Langzeitüberleben nach Herzkreislaufstillstand.

INTRODUCTION: Due to limited resources, the 2010 European Resuscitation Council (ERC) guidelines could not be fully implemented in the Emergency Medical Services (EMS) of Brunswick, Germany. This is why implementation was prioritized according to local conditions. Thus, prehospital therapeutic hypothermia, mechanical chest compression and feedback systems were not established. Clinical data and long-term results were assessed by a QM system and room for improvement was identified. METHODS: All attempted resuscitations from 2011 until 2014 were recorded and compared against the German Resuscitation Registry. Outcomes of adult patients following non-traumatic cardiac arrest were analyzed by year. RESULTS: 812 resuscitations were attempted (incidence 81.2/100,000 inhabitants/year). In the two years following full implementation since 2013 the discharge rate from hospital was 16.4 %, the discharge rate with a favorable neurologic outcome was 14.1 %, the 1-year survival rate was 14.4 % in 2013. A significant improvement of risk-adjusted ROSC rate during the investigation period was demonstrated. The discharge rates remained unchanged; the increase in the discharge rates paralleled the increase in CPR incidence. EMS response times were remarkably shorter. CONCLUSION: The implementation of the ERC guidelines chosen appears to be generally safe. Fast EMS response contributed to superior results. All links of the chain of survival showed room for improvement, especially the proportion of lay rescuer CPR and telephone-assisted CPR. The high CPR incidence might indicate room for improvement in prevention. Access to resuscitation care can hardly be evaluated. Age-related access to pre-hospital resuscitation seems to be appropriate.

Interobserver Agreement of the Echocardiographic Diagnosis of LV Hypertrabeculation/Noncompaction.

OBJECTIVES: The aim of the study was to assess interobserver agreement (IOA) between 3 observers from 2 laboratories. BACKGROUND: IOA of left ventricular hypertrabeculation/noncompaction (LVHT) in adults has only been studied within single echocardiographic laboratories. METHODS: Echocardiographic recordings with and without LVHT were selected and anonymized. The "not-LVHT" cases were matched for age and systolic function. Each observer reviewed the recordings, blinded to the initial diagnosis and the other observers' results. Pre-defined criteria for LVHT were: 1) >3 prominent trabeculae at end-diastole, distinct from papillary muscles, false tendons, or aberrant bands; 2) a noncompacted part of a 2-layered myocardial structure formed by these trabeculations; 3) a ratio of >2:1 of noncompacted to compacted layer at end-systole; and 4) perfusion of the intertrabecular spaces from the ventricular cavity. IOA was estimated using the kappa measure of concordance. RESULTS: Cine-loops of 100 patients (42 women, ages 16 to 92 years), 50 from each center, and 51 with LVHT as the initial diagnosis, were reviewed. The left ventricular end-diastolic diameter was 32 to 78 mm, and ejection fraction, 4% to 88%. The observers agreed about presence (n = 29) or absence (n = 36) of LVHT and disagreed in 35 cases. Agreement was higher among the 2 observers from the same laboratory (kappa 0.793 [95% confidence interval (CI): 0.672 to 0.915]) than from different laboratories (kappa 0.628 [95% CI: 0.472 to 0.784], kappa 0.669 [95% CI: 0.521 to 0.818]). The observers agreed with the initial report of LVHT-presence in 53% and of absence in 67%. By reviewing the discordant cases, consensus was achieved about LVHT presence (n = 8) or absence (n = 16); in 11 cases, the diagnosis remained questionable. Discordance was due to poor image quality, lack of views in different apical planes, aberrant bands and chordae tendineae, abnormally sized or inserting papillary muscles, and localized calcifications of the endocardium. CONCLUSIONS: IOA was substantial for diagnosing LVHT. However, even the application of pre-defined criteria yielded disagreement in 35% of cases; and after mutual review, there were still 11% questionable cases.

An IT Architecture for Systems Medicine.

Systems medicine aims to support treatment of complex diseases like cancer by integrating all available data for the disease. To provide such a decision support in clinical practice, a suitable IT architecture is necessary. We suggest a generic architecture comprised of the following three layers: data representation, decision support, and user interface. For the systems medicine research project "Clinically-applicable, omics-based assessment of survival, side effects, and targets in multiple myeloma" (CLIOMMICS) we developed a concrete instance of the generic architecture. We use i2b2 for representing the harmonized data. Since no deterministic model exists for multiple myeloma we use case-based reasoning for decision support. For clinical practice, visualizations of the results must be intuitive and clear. At the same time, they must communicate the uncertainty immanent in stochastic processes. Thus, we develop a specific user interface for systems medicine based on the web portal software Liferay.

Feasibility study of a sensor-based autonomous load control exercise training system for COPD patients.

Decision support systems (DSSs) which are able to automatically supervise and control physical exercise training of patients affected by chronic obstructive pulmonary disease (COPD) are regarded as a novel method to promote rehabilitation. The objective of our research work for this paper was to evaluate the feasibility of a rule-based DSS for autonomous bicycle ergometer training of COPD patients. Load control is based on real-time analysis of sensor parameters oxygen saturation and heart rate. Ten COPD patients have participated in a study, performing altogether 18 training sessions. On average, 7.4 rules were fired in each training session. Four sessions had to be stopped for different reasons. The average ergometer training load ranged between 31 and 47 W. The average percentage of heart rate in or lower than the intended target zone was 45.9 and 41.6%, respectively. The average patient-perceived Borg value was 12.6±2.4. Patients reported a high satisfaction for the automatically controlled training. With the help of the DSS, patients may change their training place from a rehabilitation center to their own homes. More studies are needed to assess long-term clinical and motivational effects of the DSS in home environment.

Information and communication technologies for promoting and sustaining quality of life, health and self-sufficiency in ageing societies--outcomes of the Lower Saxony Research Network Design of Environments for Ageing (GAL).

Many societies across the world are confronted with demographic changes, usually related to increased life expectancy and, often, relatively low birth rates. Information and communication technologies (ICT) may contribute to adequately support senior citizens in aging societies with respect to quality of life and quality and efficiency of health care processes. For investigating and for providing answers on whether new information and communication technologies can contribute to keeping, or even improving quality of life, health and self-sufficiency in ageing societies through new ways of living and new forms of care, the Lower Saxony Research Network Design of Environments for Ageing (GAL) had been established as a five years research project, running from 2008 to 2013. Ambient-assisted living (AAL) technologies in personal and home environments were especially important. In this article we report on the GAL project, and present some of its major outcomes after five years of research. We report on major challenges and lessons learned in running and organizing such a large, inter- and multidisciplinary project and discuss GAL in the context of related research projects. With respect to research outcomes, we have, for example, learned new knowledge about multimodal and speech-based human-machine-interaction mechanisms for persons with functional restrictions, and identified new methods and developed new algorithms for identifying activities of daily life and detecting acute events, particularly falls. A total of 79 apartments of senior citizens had been equipped with specific "GAL technology", providing new insights into the use of sensor data for smart homes. Major challenges we had to face were to deal constructively with GAL's highly inter- and multidisciplinary aspects, with respect to research into GAL's application scenarios, shifting from theory and lab experimentation to field tests, and the complexity of organizing and, in our view, successfully managing such a large project. Overall it can be stated that, from our point of view, the GAL research network has been run successfully and has achieved its major research objectives. Since we now know much more on how and where to use AAL technologies for new environments of living and new forms of care, a future focus for research can now be outlined for systematically planned studies, scientifically exploring the benefits of AAL technologies for senior citizens, in particular with respect to quality of life and the quality and efficiency of health care.

A prospective field study for sensor-based identification of fall risk in older people with dementia.

OBJECTIVE: Aim of this study was to make a fall prognosis in a cohort of older people with dementia in short-term (2 month), mid-term (4 month) and long-term (8 month) intervals using accelerometry during the subjects' everyday life. METHODS: The study was designed as a longitudinal cohort study. The subjects were recruited from a nursing home and geriatric assessment tests were conducted at baseline. Each subject underwent four visits and was measured at each visit for one week. Gait episodes were detected and gait parameters were extracted from these episodes. These gait parameters were combined with the falls occurred during the study. A decision tree induction method was used to analyze the data. RESULTS: Forty subjects participated in the study, whereby 12 drop-outs were registered. The geriatric assessment tests were unable to distinguish between the groups (AUC < 0.6). The evaluation of the models induced with the decision tree classification showed a rate of correctly classified gait episodes of 88.4% for short-term, 74.8% for mid-term, and 88.5 % for long-term monitoring. DISCUSSION AND CONCLUSIONS: We concluded that it is possible to classify gait episodes of fallers and non-fallers in people with dementia during everyday life using accelerometry.

A methodological framework for the analysis of highly intensive, multimodal and heterogeneous data in the context of health-enabling technologies and ambient-assisted living.

BACKGROUND: When applying health-enabling technologies (HET), researchers are faced with analyzing highly intensive, multimodal and heterogeneous data sets. Experience has shown that there is a lack of understanding concerning the relationship of analysis methods suitable for such data sets and their appropriate application. OBJECTIVES: The objective of this paper is to describe the present situation when analyzing data of HET and the main problems in this context, to present a nomenclature suitable for analysis methods in the context of HET, to present an example dealing with geriatric diseases that highlights the problems and the urgent need for results and to explain some steps for future work. METHODS: Nomenclatures as standard tools in information processing are applied. RESULTS: We present an open three-axial mono-hierarchical nomenclature called SNOCAP-HET. Moreover, we explain other ideas to overcome the lack of systematization within the set of analysis methods suitable for HET. CONCLUSIONS: Our approach allows for an extension of SNOCAP-HET and will allow for the development and evaluation of different measures for the appropriateness of analysis methods given a certain highly intensive, multimodal and heterogeneous data set in the context of HET. Our final future aim is to obtain better results when analyzing medical data.