Using Voogle to Search Within Patient Records in the VA Corporate Data Warehouse

Voogle requires that the user initiates the indexing process prior to initiating the search process. Although Voogle defaults to 3 years prior to the current date, the user can specify a start date extending to close to the year 2000. The volume of data flowing into the CDW precludes automatic indexing of all patient data, as well as automatic updating of previously indexed data. We have explored the feasibility of queueing scheduled appointments for the following day, and although the strategy shows some promise, avoiding conflict with user-requested on-demand indexing remains challenging.

The current VA network architecture updates the CDW every night, resulting in up to a 24-hour delay in data availability. However, this delay should be reduced to several minutes after implementation of real-time data feeds accompanying the coming transition to a new EHR platform.

Conclusions

The recent introduction of the Joint Legacy Viewer (JLV) to the VA EHR desktop has enhanced the breadth of patient-specific information available to any VHA clinician, with recent enhancements providing access to some community care notes from outside HCPs. Voogle builds on this capability by enabling rapid search of text notes and structured data from multiple VA sites, over an extended time frame, and perhaps entered by hundreds of authors, as demonstrated in the case example. Formal usability and workload studies have not been performed; however, anecdotal reports indicate the application dramatically reduces the time required to search for critical information needed for care of complex patients who have been treated in multiple different VA hospitals and clinics.

The Voogle paradigm of leveraging patient information stored within a large enterprise-wide data warehouse through NLP techniques may be applicable to other systems as well, and warrants exploration. We believe that replacing traditional data search paradigms that require knowledge of data structure with a true query-based paradigm is a potential game changer for health information systems. Ultimately this strategy may help provide an antidote for the information chaos impacting HCP cognition. Moreover, reducing HCP cognitive load and time on task may lessen overall health care costs, reduce provider burn-out, and improve the quality of care received by patients.

Near real-time data feeds and adding additional clinical domains will potentially provide other benefits to patient care. For example, the authors plan to investigate whether sampling incoming data may assist with behind-the-scenes continuous monitoring of indicators of patient status to facilitate early warning of impending physiologic collapse.¹⁰ Other possible applications could include real-time scans for biosurveillance or other population screening requirements.

Acknowledgments
The authors express their sincere appreciation to Leslie DeYoung for documentation and Justin Wilson who constructed much of the graphical user interface for the Voogle application and design. Without their expertise, passion, and commitment the application would not be available as it is now.