Ontology-Based Annotation of Time-Series Data

Dr. Raimond Winslow

Institute for Computational Medicine, Center for Cardiovascular Bioinformatics and Modeling, Johns Hopkins University

Cardiovascular disease (CVD) is a major cause of morbidity and mortality in the United States with almost 1 million deaths related to CVD and more than 70 million Americans with CVD. Understanding the cause and treatment of CVD requires a truly integrative approach, spanning the molecular to systems level. The Reynolds project seeks to discover new methods for identifying those patients who are at high risk for sudden cardiac death (SCD) subsequent to coronary artery disease (CAD). This is a significant problem because SCD in this population accounts for almost half of all CVD-related deaths per year. Risk is sufficiently high in patients with prior myocardial infarction and left ventricular dysfunction (LVD) that placement of implantable cardioverter-defribillators (ICDs) produces a significant reduction in mortality. The Reynolds investigators are collecting multi-scale data from patients in a large cohort diagnosed with CAD and LVD and who have received an ICD. The goal of the proposed work is to be able to reliably predict which patients with CAD are at highest risk for SCD. The Reynolds project provides an ideal dataset for the proposed tools.


The aims of this project are to develop an electrocardiogram (ECG) ontology using NCBO technology and collect community feedback via the notes feature of BioPortal. The second aim is to develop an ECG data model and data service using tools that are part of the Protégé backend. The association of ontology elements with their computed values from ECG analysis will facilitate the CV research community to share and analyze primary and derived ECG data. Due to the variety of ECG data formats used by different ECG instrument vendors, each with a proprietary data format, ECG data is almost never disseminated with any mechanism for researchers to replicate and assess the accuracy of the derived data. The third aim is to extend the capabilities of an existing ECG data management and analysis portal in an effort to replace manual annotation of paper ECG chart recordings. A parser will be developed to associate values from automated analysis algorithms with ECG ontology terms utilizing the BioPortal API to return ontology term URIs. The annotation of ECG waveforms will be implemented on the Cardiovascular Research Grid portal with term information returned to the user interface via BioPortal REST services.