SWAT4HCLS 2024 Biohackathon

Overview

SWAT4HCLS (Semantic Web Applications and Tools for Health Care and Life Sciences) is an annual conference at the intersection of semantic web technologies and biomedical research.Each year, the conference includes a dedicated Biohackathon track — a collaborative working session that runs in parallel with the main conference and continues with a dedicated full day afterwards.

I participated in the SWAT4HCLS 2024 Biohackathon in Leiden, Netherlands, contributing to discussions and collaborate across two projects.

Projects

Semantic OMOP and Alignment with HL7 FHIR

The OMOP Common Data Model (OMOP CDM) is one of the most widely adopted standards for representing observational health data. However, its relational structure poses challenges for semantic interoperability. This is particularly true when attempting to align OMOP with other health data standards such as HL7 FHIR and SPHN RDF Schema.

This project explored the challenges involved in defining a semantic representation of OMOP and how such a representation could be aligned with FHIR. Key discussion points included:

• How to faithfully represent OMOP's concept-based vocabulary system (driven by OMOP Vocabulary) in RDF/OWL while preserving its original semantics
• Where OMOP and FHIR share overlapping clinical concepts and where they diverge - and how to bridge those gaps through semantic mappings
• The trade-offs involved in choosing between a direct RDF translation of OMOP versus a higher-level ontological representation that abstracts away the relational model

The project was exploratory in nature, aimed at identifying the key friction points and laying the groundwork for a more interoperable semantic layer over OMOP-standardised datasets.

Lightweight Upper-Level Ontology for Clinical Concepts

A recurring challenge in clinical data modeling is the lack of a shared, minimal upper-level ontology that can serve as a common anchor for representing concepts from diverse clinical data models without the complexity and commitment of adopting a full foundational ontology.

This project explored a new lightweight upper-level ontology for representing core clinical concepts, with the goal of providing just enough semantic structure to enable meaningful interoperability between clinical data models such as OMOP, FHIR, and SPHN RDF Schema, while remaining approachable for implementers without deep ontology expertise.

The project mainly focused on hands on exploration of a new lightweight upper-level ontology developed in the context of the AIDAVA project from Michel Dumontier's Lab.