Current work in progress investigates metacognition and decision-making style as they are reflected in speech data provided by dermatologists.
Metacognition concerns physician self-assessment of their own skill. Physicians with what is termed metacognitive awareness tend to know what they know, and also what they do not know – that is, their confidence is a good measure of their diagnostic accuracy. Metacognitive awareness, in turn, has been shown improve decision-making.
The study attempts to determine the linguistic markers of metacognition. Do certain linguistic features, such as a lexical (e.g., word choice), or acoustic-prosodic (e.g., pitch) indicate physician level of metacognitive awareness?
In addition to linguistic markers, features of eye gaze may also provide insight into physicians' metacognition. Of particular interest is the integration of these different modalities into "multimodal" features (e.g. overlaps of eye-gaze patterns and lexical markers) in an attempt to model the interdependent nature of these phenomena.
Metacognitive awareness, as determined by linguistic and gaze markers, can then be used to characterize system users in educational applications, so that the system can respond to user over- and under-confidence, and track user progress from novice to expert.
McCoy W*., Ovesdotter Alm C., Calvelli C., Pelz J., Shi P., Haake A. (2012) Linking uncertainty in physicians’ narratives to diagnostic correctness (pp. 19-27). Proceedings of the Workshop on Extra-propositional aspect of meaning in computational linguistics at the 50th Annual Meeting of the Assoc. for Computational Linguistics 2012, Jeju, Korea.
Womack K*., Ovesdotter Alm C., Calvelli C., Pelz J., Shi P., Haake A. (2013) Markers of Confidence and Correctness in Spoken Medical Narratives. Proceedings of Interspeech 2013, Lyon, France. In print.
Bullard, J., Alm, C. O., Yu, Q., Shi, P., & Haake, A. (2014). Towards multimodal modeling of physicians' diagnostic confidence and self-awareness using medical narratives. In Proceedings of COLING, the 25th International Conference on Computational Linguistics: Technical Papers (pp. 1718-1727).