Using i2b2 to Bootstrap Rural Health Analytics and Learning Networks

Abstract

We demonstrate that the open-source i2b2 (Informatics for Integrating Biology and the Bedside) data model can be used to bootstrap rural health analytics and learning networks. These networks promote communication and research initiatives by providing the infrastructure necessary for sharing data and insights across a group of healthcare and research partners. Data integration remains a crucial challenge in connecting rural healthcare sites with a common data sharing and learning network due to the lack of interoperability and standards within electronic health records. The i2b2 data model acts as a point of convergence for disparate data from multiple healthcare sites. A consistent and natural data model for healthcare data is essential for overcoming integration issues, but challenges such as those caused by weak data standardization must still be addressed. We describe our experience in the context of building the West Virginia/Kentucky Health Analytics and Learning Network, a collaborative, multi-state effort connecting rural healthcare sites.

Fig. 1

The West Virginia/Kentucky Health Analytics and Learning Network (HALN) is a multi-state effort to establish a collaborative network of rural health clinics.

Fig. 2

We can map each hospital or clinic's data to the i2b2 data model and connect them together as a SHRINE-based network. Mapping the data allows us to leverage the existing i2b2 query tool and to build the necessary analytical meta-layer that supports consistent reporting and intelligent extraction of data across sites.

Fig. 3

Observations are central to the i2b2 data model; one patient and one visit can have many observations.

Fig. 4

Visualization rapidly reveals what clinical issues might exist by simply mapping laboratory results with respect to their known ranges.

Fig. 5

Visualization quickly reveals outlying values and can assist in finding insights per semantic groupings.