A visual interface designed for novice users to find research patient cohorts in a large biomedical database

Abstract

One of the more difficult tasks of informatics is allowing for the navigation of complex databases. At Partners Healthcare Inc. we have developed an analytical database to allow for searching clinical data to obtain cohorts of patients for research studies. The characteristics of the patients within the cohorts must often comply with complex inclusion and exclusion criteria. The users of the database are research clinicians, often with no prior database experience. To assist these clinicians in finding their patient cohorts, we constructed a Querytool that they use directly to find their desired populations. In order to understand if the Querytool could indeed be used successfully by novice users, we analyzed the first 10 queries of 219 users. This analysis was able show that novice users are able to achieve excellent success using the Querytool

Figure 1

A hierarchical tree of items for users to choose from is present in the pane along the left, and a set of panels to roughly model a Venn-diagram are present in the upper right. The “New group” button (circled) creates new query-panels.

Figure 3

The “Previous Queries” tab is selected on the left large panel in this figure. Also shown are the “Value” options, which are described further in the text. Note that units for the values may be selected in the lower drop-down selector.

Figure 4

Screen created to fix timing errors.

Figure 2

The “Find” tab (circled) can be used to access a function that will alphanumerically search the vocabularies. Also shown is the “Constraints” options which allows dates and special characteristics associated with the semantic type (in this case diagnoses) to be chosen. Available for diagnoses are primary/secondary/admission status of the diagnosis, and an option to search based only upon the parent code (and not the children). Also circled (towards the right) is the timing option selector.