UNRAVEL: big data analytics research data platform to improve care of patients with cardiomyopathies using routine electronic health records and standardised biobanking

Abstract

Introduction: Despite major advances in our understanding of genetic cardiomyopathies, they remain the leading cause of premature sudden cardiac death and end-stage heart failure in persons under the age of 60 years. Integrated research databases based on a large number of patients may provide a scaffold for future research. Using routine electronic health records and standardised biobanking, big data analysis on a larger number of patients and investigations are possible. In this article, we describe the UNRAVEL research data platform embedded in routine practice to facilitate research in genetic cardiomyopathies.

Design: Eligible participants with proven or suspected cardiac disease and their relatives are asked for permission to use their data and to draw blood for biobanking. Routinely collected clinical data are included in a research database by weekly extraction. A text-mining tool has been developed to enrich UNRAVEL with unstructured data in clinical notes.

Preliminary results: Thus far, 828 individuals with a median age of 57 years have been included, 58% of whom are male. All data are captured in a temporal sequence amounting to a total of 18,565 electrocardiograms, 3619 echocardiograms, data from over 20,000 radiological examinations and 650,000 individual laboratory measurements.

Conclusion: Integration of routine electronic health care in a research data platform allows efficient data collection, including all investigations in chronological sequence. Trials embedded in the electronic health record are now possible, providing cost-effective ways to answer clinical questions. We explicitly welcome national and international collaboration and have provided our protocols and other materials on www.unravelrdp.nl .

Keywords: Big data analytics; Biobanking; Cardiomyopathy; Electronic health record; Machine learning; Research data platform.

Fig. 1

Schematic overview of different types of included data. In short, data on investigations and metadata are automatically extracted after informed consent has been provided. Additionally, patient demographics and specific events, such as date of admission, are included. Information from the municipality registry can be requested concerning, for example, death

Fig. 2

Temporal character of health care data. Schematic overview of a temporal window in which patients visit the centres. In contrast to manually maintained registries where data may be disregarded, the UNRAVEL research data platform includes all (meta)data and investigations. ECG electrocardiogram, MRI magnetic resonance imaging, Hb haemoglobin, BNP brain natriuretic peptide, CRP C-reactive protein

Fig. 3

Two data tables from the UNRAVEL research data platform as samples of electrocardiogram and echocardiogram output in SAS enterprise guide. ECG electrocardiogram, ECH echocardiogram

Fig. 4

Sample data from the text-mining tool, where based on the clinical notes in the electronic health records (DECURSUS) an output file is created with different standardised variables, such as arterial hypertension, diabetes and dyslipidaemia. Variables are harmonised with the German TORCH registry, but can be changed as deemed necessary. The application is written for Dutch cardiovascular notes