Free text makes it convenient for clinicians to capture rich information about a patient in narratives and first-person stories. But this makes EMRs hard to analyze for big data applications. Despite advances in Natural Language Processing methods, building models is expensive and time-consuming. It requires a long collaboration between clinicians and data-scientists. Clinicians provide annotations and training data, while data-scientists build the models. Further, the current approaches do not provide for clinicians to inspect these models and give feedback. This forms a barrier to NLP adoption in the clinical domain, resulting in fewer real-world applications. Building models interactively can help narrow the gap between clinicians and data-science. Interactive learning systems could allow clinicians to build NLP models on their own without machine learning experience, while also reducing the need for prior annotations. Such systems may provide users with a review, feedback, and retrain loop to build models that can be revised with every iteration.
In my dissertation work, I design, build and evaluate prototype systems for both clinical care and research applications. I demonstrated this approach for training binary classification models in NLPReViz to help with retrospective review of clinical notes. I am currently working on an intelligent tool that can highlight important parts of clinical notes for critical care physicians. Building these prototypes serve as a demonstration of my approach and help understand the design of such systems for wider clinical NLP applications. I follow a two-part evaluation scheme for my prototypes: a) a usability study with clinicians to demonstrate feasibility and usefulness, followed by b) an empirical evaluation to evaluate model correctness and utility as the physicians use them to solve real-world problems.
After participating in this activity, the learner should be better able to:
- Understand interactive system development
- Discuss the use of NLP for clinical work
- Understand usability studies of interactive NLP development
Gaurav Trivedi, MS, B.Tech
University of Pittsburgh
Gaurav is a PhD Candidate in Intelligent Systems Program (Artificial Intelligence), School of Computing and Information at University of Pittsburgh. His dissertation work is on designing interactive methods for natural language processing on clinical records. He is working with Dr. Harry Hochheiser from the Department of Biomedical Informatics, along with a group led by Dr. Shyam Visweswaran on an NIH funded project on “Development and Evaluation of a Learning Electronic Medical Record System." Previously, he worked in collaboration with the NLP group: Dr. Janyce Wiebe, Dr. Rebecca Hwa and Dr. Wendy Chapman on an NIH funded project on interactive review and revision of NLP models for retrospective research on clinical notes. He holds a Master of Science degree from University of Pittsburgh and an engineering bachelor’s degree in Information Technology from National Institute of Technology Karnataka, Surathkal. He worked with Nvidia before joining graduate school.