Work

The Development and Evaluation of a Learning Electronic Medical Record System

Year: 2018
Type: article
Abstract: Electronic medical record (EMR) systems are capturing increasing amounts of data per patient. For clinicians to efficiently and accurately understand a patient’s clinical state, better ways are needed to determine when and how to display patient data. The American Medical Association envisions EMR systems that manage information flow and adjust for context, environment, and user preferences. We developed, implemented, and evaluated a prototype Learning EMR (LEMR) system with the aim of helping make this vision a reality. A LEMR system, as we employ the term, observes clinician information seeking behavior and applies it to direct the future display of patient data. The development of this system was divided into five phases. First, we developed a prototype LEMR interface that served as a testing bed for LEMR experimentation. The LEMR interface was evaluated in two studies: a think aloud study and a usability study. The results from these studies were used to iteratively improve the interface. Second, we tested the accuracy of an inexpensive eye-tracking device and developed an automatic method for mapping eye gaze to patient data displayed in the LEMR interface. In the two studies we showed that an inexpensive eye-tracking device can perform as well as a costlier device intended for research and that the automatic mapping method accurately captures the patient information a user is viewing. Third, we collected observations of clinician information seeking behavior in the LEMR system. In three studies we evaluated different observation methods and applied those methods to collect training data. Fourth, we used machine learning on the training data to model clinician information seeking behavior. The models predict information that clinicians will seek in a given clinical context. Fifth, we applied the models to direct the display of patient data in a prospective evaluation of the LEMR system. The evaluation found that the system reduced the amount of time it takes for clinicians to prepare for morning rounds and highlighted about half of the patient data that clinicians seek.
Language: English
Open Access status: green
OpenAlex ID: W2900928120
FWCI: 0.218
Cites: 92
Cited by: 2
Related to: 10