Development and evaluation of a pulmonary graphical display

Critical events are common in the operating room. In an attempt to improve the anesthesiologist's situational awareness, I have designed a graphical display that uses color, texture, shape and emergent features to highlight abnormal pulmonary physiology. The graphical pulmonary display was developed using an iterative development cycle. With each development cycle, the users were asked to label elements of the graphical pulmonary display. The subjects' labels were compared to the designer's intent. Differences influenced the design changes. After five design iterations, anatomic intuitiveness increased by 25%, variable mapping intuitiveness increased by 34%, and diagnostic accuracy decreased by 4%. To evaluate the pulmonary graphic display's utility, 19 anesthesiologists were asked to participate in a high resolution simulator-based evaluation. Each subject was trained for 10 minutes on the pulmonary display. Half the subjects used the pulmonary graphical display and half did not while identifying and treating 5 critical pulmonary events. An obstructed endotracheal tube was diagnosed 114 seconds faster (87 ± 49 seconds vs 201 ± 68 seconds, p < 0.05) and treated 127 seconds faster (111 ± 43 seconds vs. 238 ± 60 seconds, p < 0.05) when the pulmonary display was used. Intrinsic PEEP was diagnosed earlier (160 ± 109 seconds vs 277 ± 38 seconds) and treated faster (189 ± 111 seconds vs. 277 ± 38 seconds) with the pulmonary display compared to the control condition. Next I measured the acceptance of the pulmonary graphical display in an intensive care unit. Physicians, respiratory therapists, and nurses looked at the pulmonary graphical display an average of six, three, and one times, respectively, per patient room entry. The pulmonary graphical display was perceived as useful, a desirable addition to current ICU monitors, and an accurate representation of respiratory variables. Subjects liked the simplicity of the design. This dissertation is a compilation of three publishable papers. The first paper was accepted and published by the Journal of American Medical Informatics Association. The second paper has been submitted to Anesthesia and Analgesia and is currently under review. The third paper has been accepted by the Journal of Bioinformatics and will be in publication spring, 2005.