Can Machine Learning Help us Improve Physician Communication?: Byron Wallace, Brown University (Data Management Seminar)
- Starts: 11:00 am on Friday, November 8, 2013
- Ends: 12:00 pm on Friday, November 8, 2013
Abstract Physician-patient communication is a critical component of health-care. Several studies have reported an association between metrics of physician-patient communication quality and health outcomes, and there is evidence that the relationship between the physician and the patient affects patient satisfaction and burden of symptoms. Recognizing its importance, health sciences researchers have investigated clinical communication at length, but this work has been predominantly qualitative in nature. There is a pressing need to better understand clinical interaction processes quantitatively. To this end, researchers have recently introduced coding schemas that annotate the utterances in transcribed physician-patient interactions with codes that capture clinically meaningful properties of speech. Modern schemas are multidimensional: they capture both subject matter (discussion topics) and interaction processes (speech acts). Analyzing transcripts coded with such schemas (coupled with health outcomes data) may reveal important, reproducible properties of physician communication that correlate with outcomes of interest. But two major limitations preclude progress on this front: (1) fine-grained coding of transcripts is laborious and expensive, limiting sample sizes and hence the set of questions that can be addressed; and (2) we lack models suitable to the multi-dimensional, sequential nature of physician-patient interactions. Machine learning approaches can potentially help solve both problems. In this talk I will present emerging work toward this end. Specifically, I will present a generative model of clinical communication that jointly captures both the topics and the speech acts used in each utterance. I will also present some preliminary results leveraging output from this model, and I will conclude by discussing future directions for this work. Bio Byron Wallace is an assistant professor (research) in the Department of Health Services, Policy & Practice at Brown University; he is also affiliated with the Brown Laboratory for Linguistic Processing (BLLIP) in the department of Computer Science. His research is in data mining/machine learning and natural language processing with an emphasis on applications in health informatics. Before moving to Brown, he completed his PhD in computer science under the supervision of Carla Brodley. He was awarded the Tufts Outstanding Graduate Researcher at the Doctoral Level award in 2012 and was selected as the runner-up for the 2013 ACM SIGKDD Doctoral Dissertation Award for his thesis work, which concerned developing novel machine learning methods to make conducting biomedical systematic reviews more efficient.
- MCS 148