Priming intelligent split menus with text corpora for computerized patient record data-entry

This paper describes a methodology for using the knowledge in existing health care text corpora to prime intelligent split menus for provider data-entry. A split menu is one where the top portion of a menu list is organized by user-selection frequency and the bottom portion of the list is traditionally organized in alphabetical order. A simulation shows that data-entry with these intelligent split menus requires between two and five times less effort (in terms of user selections or mouse clicks) than menus arranged alphabetically. This paper uses a corpus from echocardiography to develop the simulation. The methodology uses statistical associations between word categories in the corpus such as 'anatomy' or 'pathology' to prime the frequency ordering of the menus. A dictionary of terms contains the categorical information. After the initial priming, actual user selections are used to update the frequencies used to adapt to providers' individual data-entry patterns.