Abstract Body: What do Data Models do?
Establishing interoperable health data-sharing platforms for patients, care providers, communities, and researchers is immensely instrumental for transforming the health care system into a continuously learning and improving healthcare model [[i]]. For this purpose, several federated distributed networks have been formed among organizations to allow researchers to access national resources and create a network that can distribute queries to each participating organization in a privacy compliant manner. Among the other technical challenges, the data model agreement and its proper version control are directly attributable to the success of creating a functioning federated network. Despite the existence of well-known, well-curated Common Data Models (CDMs) such as Observational Medical Outcomes Partnership (OMOP) and Patient-Centered Clinical Research Network CDM, we would like to exercise developing a relatively small data models for potential use on an intra institutional or potential extension to the existing CDMs. In principle, CDMs define data definitions to include as well as the data representation once imported. While the promise of a data model is to allow differing entities to seamlessly compare and query data, the proper mapping, evaluation and ability to incorporate new and changing data sets decide its usability and scalability.

Hackathon Format:
Groups of 4-5 people will address common/known challenges of creating data models (e.g. loss of specificity, ambiguous mapping, future-proofing) using Patient Reported Outcomes (PROs) as a working example. Under guidance, each group will be asked to develop a prototype solution for a custom PRO survey, then groups will work to define a data model that can comprehensively represent the collection of questions and responses.

To address these common needs, the goals for the Data Model Hackathon are to learn how to:

1. Recognize hurdles in data collection and Extract Transform and Load processes
2. Adapt a model or extend an existing one to allow new data
3. Extract data that represents the intended meaning and is also easily analyzable

Each group will need to have at least one laptop to participate in hands on aspect. No specific programming expertise is needed.


Brian Ostasiewski (Presenter)
Wake Forest Baptist Medical Center

Terra Colvin Jr. (Presenter)
Wake Forest Baptist Medical Center

Umit Topaloglu (Presenter)
Wake Forest Baptist Medical Center