An OMOP Common Data Model pipeline for electrocardiography dataset integration by standardizing signal metadata and...

Efficient data sharing is crucial for the secondary use of healthcare data, which represents one of the greatest opportunities to advance pathophysiological knowledge and improve health outcomes. However, data volume and the variety of formats and standards significantly hinder data interoperability, compromising multicentric studies. Enhancing data FAIRness can increase the volume of well-characterized data to train more generalizable, interpretable, and multimodal machine learning models. The extraction and management of high-quality metadata can help institutes describe and represent their data in heterogeneous and unstructured datasets comprising biosignals, bioimages, and omics data. Metadata provides ready-to-use descriptors, enabling researchers to filter the datasets to identify samples of interest for their analyses. The Observational Medical Outcomes Partnership Common Data Model (OMOP CDM) provides a flexible solution for storing metadata from multiple healthcare data types, using unique concepts to ensure effective integration across data sources and semantic interoperability. This work presents an OMOP CDM-based standardization pipeline to extract and enable coherent representation of electrocardiographic (ECG) signal features, including arrhythmias and heart rate variability (HRV) indices, starting from public ECG datasets with different characteristics. We structured the extracted information into OMOP CDM tables and mapped each feature to its closest concept in the standard vocabularies. We found suitable standard concepts for all the ECG annotations and some HRV features. For those features that could not be mapped to standard concepts, we illustrate a workflow to establish special, OMOP-compliant mappings. We discuss the pros and cons of these approaches and propose strategies to improve the management of data standardization pipelines and custom vocabularies in federations. To help researchers test our data integration approach, open-source code covering the entire pipeline is provided. OMOP CDM proved a valid tool for harmonizing ECG metadata. However, its standard vocabularies require extensions to allow describing outputs from advanced signal processing techniques.