INCA: interaction analytics for automatic assessment of communication quality in primary care

Communication between physician and patient is crucial to the overall quality of primary care. An important element of this interaction is the physician-patient interview. While there are various opinions on what constitutes effective communication in a medical interview, sometimes supported by formal (qualitative and quantitative) studies, empirical evidence in this area is relatively scarce.
Part of the difficulty in studying interaction in medical interviews is that, in order to analyse verbal and non-verbal aspects of interaction which might influence patient outcomes, one must annotate audio or video recordings of interviews in detail. In published literature, this has mostly been done manually. As this process is time consuming and error-prone, the scale and value of such studies is limited.
This project will investigate the feasibility of a technological mechanism for electronic gathering and automated analysis of physician-patient interaction during medical interviews. It will apply state of the art technology in speech processing, text analytics and social signal processing, and investigate the impact of models through which comprehensive, data-intensive communication analysis could be conducted. This interaction analytics research will use routinely collected audio-visual data from consultations between patients and trainee general practitioners.
This research project involves automatic speech and text processing. A research outcome will be the error measures of automated speech transcription and categorisation. The non-verbal behaviour of clinicians will be examined. Outcomes include variation by clinician and by section of the medical interview, and quantification of the frequency of each form of non-verbal behaviour. Interview sections will be specified based on the Calgary Cambridge Guide to the Medical Interview. Non-verbal behaviour includes attitudes (empathy), and social signals such as gaze, body posture, facial expressions and gestures. Computational analysis will identify dimensions of physician verbal behaviour (words) predictive of effective communication, and communication characteristics shall be displayed visually for each physician.

Award Date
23 October 2015
Award Value
Principal Investigator
Professor Charles Normand
Host Institution
Trinity College Dublin
Health Research Awards