Our informatics students made incredible contributions to the Collaborator Showcase at this year’s OHDSI 2025 Global Symposium, presenting advanced findings that are pushing the boundaries of observational research. These future leaders showcased their work across methodology, data quality, and cutting-edge data integration.
Robert Barrett was selected to present a lightning talk titled “Improving VSAC to OMOP Mapping Using LLM-Assisted Curation,” showcasing how Large Language Models (LLMs) can help address a persistent challenge in data standardization. He also presented a poster that explored advanced text mining techniques for scientific literature, “Thematic Classification of Articles Using Graph Representations.” Hannah Lee contributed a seminal poster developing “A Framework for Understanding Bias and Real-World Clinical Electronic Health Record Data” to improve the reliability of observational studies. She also contributed to foundational work in building and characterizing the global OHDSI Evidence Network. Christelle Xiong added to the AI-driven data extraction work with her presentation, “AgentDose: Toward Accurate and Scalable Steroid Dose Extraction in OMOP Using NLP Parsers and LLM Agents.”
In the realm of multimodal data, Jen (Wooyeon) Park and Carrie (Qingrui) Wang presented research. Park’s work on the “Real-World Implementation of the Medical Imaging CDM: An Alzheimer’s Disease Use Case” aims to seamlessly integrate complex DICOM imaging data into the OMOP framework. Wang contributed to “Anatomical Location Auto-check and Standardization for Medical Images,” tackling the essential step of verifying and standardizing the location of observations within imaging studies. Gabriel Salvador focused on reproducibility, leading the work “Replicating Alzheimer’s Research Using Standardized Phenotyping with the OMOP Common Data Model Imaging Extension.” Seohu Lee authored “Toward Accurate Identification of Fontan and TGA in OMOP CDM: Registry-Anchored Algorithm Validation,” addressing the need for validated algorithms for complex congenital heart conditions.
Addressing critical infrastructure needs, Katy Sadowski was a key contributor to “Dqdb: Continuous Data Quality Testing for OMOP ETL,” an open-source tool ensuring data integrity for network research. Additionally, Maria Sanchez co-authored “Identifying Oral Health Concepts from Previously ‘OMOP-ified’ Data,” highlighting efforts to expand the OMOP model’s utility into specialized healthcare domains. Together, these students are proving that advanced analytical tools and robust data standards are key to unlocking the power of global real-world health data.
Their posters can be found online at the 2025 Global Collaborator Showcase – OHDSI.