Is OMOP CDM the right fit for your organization's health data?
Is your organization struggling to develop a data model for multi-modal health data? Are you unsure whether the OMOP CDM is the right solution for your organization at this moment? Â An OMOP-based custom data model might be the right path for you, let us show you why and how.
What is the OMOP CDM and why do organizations use it?
The Observational Medical Outcomes Partnership Common Data Model (OMOP CDM) is used to standardize healthcare data from various sources, enabling large-scale observational research and real-world evidence generation. In the more than 10 years since its inception, the OMOP CDM has been adopted by a large number of organizations across the globe: boasting over 2500 collaborators and millions of patient records across more than 83 countries [1]. The Observational Health Data Science and Informatics (OHDSI) community continues to grow since its foundation in 2014 [2]. It is widely used in regulatory, academic, and industry research for evidence-based decision-making, and its impact can be seen in the hundreds of OMOP-related research papers published since 2010 [3], which involve practical usages of the OMOP CDM, reviews, evaluation, and extensions of the model [4].
Not ready for OMOP CDM yet? A custom model can be your first step
In this article, we will share our experience in tailoring a path to the OMOP CDM by developing a custom data model for your organization that serves as an intermediary stage between the structure of your health data and the OMOP CDM. In this manner we assist you in bridging the divide between your data model and one that has been developed and broadly embraced by a growing global community of experts and help you take a first step into your OMOP journey.
Tailoring OMOP CDM to unique research and data needs
One of the key features of the OMOP CDM is its adaptability. It offers ongoing releases of new versions, modules, expansions and vocabularies to meet the evolving needs of the community and researchers. One such example is the oncology extension, which has helped cancer researchers use the OMOP CDM for cancer research [6], including AI-driven cancer prediction [6]. However, there are scenarios where the OMOP CDM may require tailoring to align with specific organizational requirements. For example, in the case of some rare diseases, for which the OMOP CDM does not have the appropriate catalog of terminologies [7]. Think of such cases where concepts for rare diseases are not yet present in OHDSI’s standard vocabularies, or in cutting-edge research, where data scientists may seek to develop a custom data model using the OMOP CDM as a starting point. This later approach allows for further finetuning and manipulation of the OMOP CDM to better suit the specific research and organisational needs. These custom OMOP-based models offer a valuable resource for organizations looking to leverage the collective expertise of the OHDSI community as a structural foundation, while implementing deviations that would otherwise not fall within the broader scope and objectives of the OHDSI community. While this approach would result in a data model that cannot leverage the OHDSI tools and collaboration with the rest of the community, it would allow organizations to have an operational model that closely aligns with the organization’s mentality and internal research needs.
A custom model that fits your needs while aligning with OMOP standards
The overarching goal of such an approach is to develop a data model customized to tightly fit with your organization’s operational needs and way of thinking and which also provides an intermediate step between the current structure and objectives of your health data and the OMOP CDM. This helps you bridge the gap between your data model and one that has been developed and broadly embraced by a global community of experts.
How to bridge the gap between multi-modal health data and the OMOP CDM?
Start with a data analysis and stakeholder input to build a tailored modelÂ
The starting point is to conduct an extensive analysis of your health data and engage in comprehensive consultation with all stakeholders to develop and implement a customized data model that you have complete control over which also aligns closely with the OMOP CDM structure.
How The Hyve supports custom OMOP-based data model development
At The Hyve, we possess extensive experience in data modeling and working with the OMOP CDM. Additionally, we have expertise in developing OMOP-based custom models, specifically tailored to the unique requirements of an organization. We develop these models with meticulous attention to detail, considering the complex nuances of any deviations or extensions to the OMOP CDM. This ensures that no obstacles arise in the future as the model or its objectives evolve, or when developing an ETL to harmonize data in the custom model with those from the OHDSI community.
Our process: from workshops to a fully functional custom data model
To achieve this, we initiate a bidirectional knowledge transfer: through a series of interviews and workshops, we ensure that everyone at The Hyve is intimately familiar with the specific needs and objectives of your organization. Simultaneously, we share our expertise and knowledge of the OMOP CDM and data modeling. Subsequently, we utilize the OMOP CDM as a foundation and commence a systematic examination of the model to ascertain which components are essential, unnecessary, or necessitate modifications to accommodate the organization. This may entail removing or restructuring tables, employing vocabularies and concepts distinct from those utilized in the OMOP CDM, incorporating new tables and fields to differentiate and organize the source data in a different manner, or even utilizing a distinct type of database. Through comprehensive consultations with the stakeholders and end-users, we ultimately establish a custom model that is suitable for the organization in a manner that is much faster and cheaper than developing a custom model from scratch: instead we start from the OMOP CDM, a strong foundation that we at The Hyve are intimately familiar with. The organization then has complete autonomy over its use and future updates.
What specific outcomes can you expect from such an approach?
What you get: custom data model, documentation, SOPs, and future OMOP compatibilityÂ
The result of such a project is a custom data model that uses the OMOP CDM as its foundation and is adapted to the needs and requirements of your users. This custom data model will already take into account all the intricacies of the source data and will be accompanied with comprehensive documentation and Standard Operating Procedures (SOPs), enabling future team members to quickly familiarize themselves with the model and its use. Furthermore, this documentation will detail all the deviations from the OMOP CDM, along with the rationale behind them. This will allow your organization to easily maintain and update the model as needed, and even develop an Extract Transform Load (ETL) pipeline to convert the data from the custom model into the OMOP CDM at a later point.
Benefits: a flexible and future-proof approach to health data standardization
If your organization is struggling to develop a data model for multi-model health data and you are unsure whether the OMOP CDM is the right solution at this moment, then this is a valuable path forward for you. We at The Hyve will help you quickly establish an OMOP-based custom data model that will solve your current operational and internal research needs by leveraging our expertise with the OMOP CDM, and simultaneously help you explore if and how the OMOP CDM and the OHDSI community can benefit your organization in the future.
ReferencesÂ
Â
[1] OHDSI. OHDSI Who We Are. https://ohdsi.org/who-we-are/
[2] OHDSI, The book of OHDSI Observational Health Data Sciences and Informatics, 2019
[3]OHDSI. OHDSI Publications. https://www.ohdsi.org/publications/
[4]Reinecke, I., Zoch, M., Reich, C., Sedlmayr, M., & Bathelt, F. (2021). The usage of OHDSI OMOP–a scoping review. German Medical Data Sciences 2021: Digital Medicine: Recognize–Understand–Heal, 95-103.
[5]Wang, L., Wen, A., Fu, S., Ruan, X., Huang, M., Li, R., Lu, Q., Williams, A. E., & Liu, H. (2024). Adoption of the OMOP CDM for Cancer Research using Real-world Data: Current Status and Opportunities. medRxiv : the preprint server for health sciences, 2024.08.23.24311950. https://doi.org/10.1101/2024.08.23.24311950
[6]Ahmadi, N., Peng, Y., Wolfien, M., Zoch, M., & Sedlmayr, M. (2022). OMOP CDM can facilitate data-driven studies for cancer prediction: a systematic review. International journal of molecular sciences, 23(19), 11834.
[7]Zoch, M., Gierschner, C., Peng, Y., Gruhl, M., Leutner, L. A., Sedlmayr, M., & Bathelt, F. (2021). Adaption of the OMOP CDM for Rare Diseases. Studies in health technology and informatics, 281, 138–142.