In the Hospital, my world was one of patient histories, differential diagnoses, and treatment plans. The data was rich, nuanced, and deeply human. But I saw a frustrating paradox: the most critical information was often locked away in unstructured notes and siloed in electronic health record (EHR) systems, making it nearly impossible for me to leverage the same data I was generating. A brief visit to the IT department confirmed that this was not only me as an individual contributor, this was a systemic problem.
This disconnect felt especially stark coming from previous work in other regulated sectors. Rigorous, systematic data analysis is not an afterthought—it’s fundamental to operations, auditing, and safety. Clinical practice, for all its importance, often lacks this data-driven feedback loop. Insights are gleaned from randomized controlled trials or manual audits, but the rich, continuous data stream from routine care goes largely unanalyzed.
Bridging this gap between clinical practice and data-driven insight became the mission behind IOMED. We set out to build the infrastructure that would allow healthcare systems to systematically learn from the vast amount of data they generate every day.
The Chaos of Clinical Data
Anyone who has worked with hospital data knows the challenge. A single concept, like “Type 2 Diabetes,” can be represented by dozens of different local codes, abbreviations, or free-text descriptions across different institutions. One hospital uses ICD-9, another uses ICD-10, and a third might have its own internal terminology.
Running a simple query like “How many patients with Type 2 Diabetes were prescribed metformin in the last year?” across just two hospitals becomes a monumental task of manual mapping and guesswork. Scaling that to ten, or a hundred, is impossible. Without a common language and structure, there can be no systematic analysis.
Finding a Structure for the Lingua Franca
The solution required two components: a common vocabulary and a common structure. The lingua franca of clinical terminology already exists in standards like SNOMED-CT, which provides a comprehensive, machine-readable vocabulary for clinical terms. But a vocabulary alone is not enough. You also need a grammar—a consistent way to structure that vocabulary to represent a patient’s journey.
This is where the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM) comes in. Maintained by the global OHDSI community, the OMOP CDM provides the structural framework that makes SNOMED-CT and other coding efforts accessible and interoperable. It organizes the data into a standard set of tables (e.g., person, condition_occurrence, drug_exposure), creating a unified schema that can be queried systematically.
OMOP is the universal translator for the structure of the data, allowing us to map disparate sources into one coherent, person-centric model.
Why We Built IOMED on OMOP
Adopting the OMOP CDM was one of the most critical foundational decisions we made. Here’s why:
- Systematic Analysis: By transforming disparate data into a single, coherent format, OMOP makes it possible to write one query that can run across an entire network of hospitals. It turns an impossible task into a routine one.
- Community and Tools: OHDSI provides a rich ecosystem of open-source analytical tools designed to work with the OMOP CDM out of the box. We weren’t just adopting a data model; we were joining a global community dedicated to advancing health analytics.
- Scalability and Future-Proofing: As our network grows, every new hospital we onboard adds to the collective power of the network without adding exponential complexity. The model is robust, continuously updated by experts, and designed for the very purpose we needed: large-scale, multi-site research.
Bridging Medicine and Data Science
For us, OMOP is more than a technical choice; it’s the bridge between the clinical world and the data-driven future of medicine. It allows us to convert the messy, fragmented reality of healthcare data into a clean, structured asset ready for advanced analytics and machine learning.
This standardization is the non-negotiable first step. It’s the foundational work that makes everything else—from identifying patient cohorts for clinical trials to discovering new treatment patterns—possible. We are laying the groundwork, one hospital at a time, for a future where real-world evidence is not a niche academic pursuit but an integral part of improving patient outcomes everywhere.