Machine Learning

Development of Computational Methods for Evaluating Doctor-Patient Communication

August 2017 - August 2020
Sponsor: Patient-Centered Outcomes Research Institute (PCORI)

Can we train the computer to study patient-physician communication? The answer is YES. See our paper in JAMIA. We conclude that statistical natural language processing, or NLP, can make studying conversations between clinicians and patients faster and easier. We use computers to look for meaning in conversations by looking at words and phrases and how they are used. NLP can define the issues discussed, describe speakers’ emotions.

Additional details about this study can be found here.

Research collaborators:

Zac Imel, PhD (PI)
Ming Tai-Seale, PhD (Co-PI)
Padhraic Smyth, PhD