AVML 2012:Daniel Angus - Generating Visual Insights into Effective Doctor-Patient Consultations

• Report rights infringement
• published: 02 Dec 2013
• views: 315

Daniel Angus (School of Information Technology and Electrical Engineering, School of Journalism and Communication, The University of Queensland) Effective communication between healthcare professionals and patients is critical to patient health outcomes. The doctor/patient dialogue has been extensively researched from different theoretical perspectives, with findings emphasising a range of communication behaviour that leads to effective communication. In this study we applied the Discursis visualisation technique [1] to analyse examples of training and clinical medical discourse transcripts. Discursis automatically builds an internal language model from a conversation transcript, tags conversation turns based on their conceptual content, and generates an interactive visual representation of the discourse under study. The analysis afforded by Discursis is useful in examining whole consultation scale interactions, and findings from this work are helping to highlight effective consultation techniques including accommodative, engagement and repetition behaviours. A significant advantage of Discursis over alternative visualisation techniques is the ability to visualise topic usage patterns across a range of time scales simultaneously. In a recent study Angus et al. [1] performed an analysis of conversations from two Australian television talk shows to highlight topic usage patterns including topic convergence (on a turn-by-turn time scale) between the participants, and whole conversation range call-back behaviours where early mentioned topics were revisited much later in the conversation. The study indicated that as a decision support tool, a discourse analyst could use the system to confirm pre-held hypotheses about the type and magnitude of interaction (in terms of topic reuse) between conversation participants, and as a forensic tool to discover patterns of interaction and interesting time periods where conversation participants demonstrated topic convergence characteristics. Findings from the application of Discursis analysis to doctor/patient consultation transcripts suggest that particular visual patterns of topic use are present in effective doctor/patient consultations. One feature of interest is what we term the 'introductory stripe' which is a band of topic recurrence stemming from the patient which occurs early in the consultation. An example of an introductory stripe is indicated in Fig. 1a. The introductory stripe is present as the patient mentioned a number of prominent topics in this single early occurring utterance. The conceptual content of this early utterance recurs with the patient's own statements (red squares) and the doctor's statements (red & blue squares) throughout the remainder of the consultation. The introductory stripe has high importance given that these concepts are repeated at multiple time-points throughout the remainder of the consultation. In essence this early turn frames much of the later occurring discussion. Platt and Gordon [2] speak of the importance in clarifying the patient's agenda, and that the doctor should try early in the interview to obtain a complete list of the patient's concerns. The 'introductory stripe' recurrence pattern is indicative of such a clarification. As a contrasting example, there is no introductory stripe present in poor consultations (Fig. 1b); instead the Doctor's early turns are the ones that recur throughout the remainder of the consultation suggesting that the Doctor provided the framing instead of the patient. References: [1] Daniel Angus, Andrew Smith, and Janet Wiles. Conceptual recurrence plots: Revealing patterns in human communication. IEEE Transactions on Visualization and Computer Graphics, 2011. in press. [2] Frederic W. Platt and Geoffrey H. Gordon. Field guide to the difficult patient interview. Lippincott Williams & Wilkins, 2nd edition, 2004.