Leveraging Eye Tracking to Prioritize Relevant Medical Record Data: Comparative Machine Learning Study.

Area of Science:

• Medical Informatics
• Human-Computer Interaction
• Machine Learning

Background:

• Electronic Medical Record (EMR) systems present vast patient data without prioritization, potentially causing physician cognitive overload.
• Learning EMR (LEMR) systems aim to mitigate this by prioritizing data relevant to the user, clinical task, and patient case.
• Supervised machine learning models are used to identify relevant data, but manual annotation for training is time-consuming and costly.

Purpose of the Study:

• To propose and evaluate eye tracking as a high-throughput, automated method for acquiring physician information-seeking behavior data.
• To train machine learning models for LEMR systems using eye-tracking data.

Main Methods:

• Critical care physicians reviewed ICU patient cases using a study-specific EMR interface.
• Physicians manually identified relevant data, while eye tracking captured gaze dwell times on data items.
• Manual annotations and gaze data were used to train and compare supervised machine learning models.

Main Results:

• 68 pairs of manual selection and gaze-derived machine learning models were developed and evaluated.
• Performance comparison using the area under the receiver operating characteristic curve showed no significant difference between manual and gaze-derived models (P=.40).

Conclusions:

• Eye tracking provides an effective, automated method for capturing physician information-seeking behavior to train LEMR systems.
• Models trained with eye-tracking data performed similarly to those trained with manual annotations.
• This supports the further development of eye tracking for training clinical decision support systems that prioritize medical data.