Abstract: In this workshop, we will demonstrate the use of Jupyter Notebooks to teach data science topics without requiring computational proficiency of either the instructor or the students. We will demonstrate this with the following use case of clinical natural language processing:
A healthcare delivery system has asked us to determine whether use of opioids is associated with increased risk of community-acquired pneumonia in adults. We have one week to identify patients with pneumonia so that the health services team can assess the risk based on pharmacy data. There is not enough time to manually read through the thousands of charts, so we will apply natural language processing for the task. Student teams in this course will collaboratively build an NLP application for identifying patients with pneumonia from a publicly available dataset. We will evaluate the predictive performance of the tools we build, compare team scores, and discuss challenges and tips for best performance.
Participants will form teams of 2-3 people, will build an NLP tool, will evaluate its performance on a training and a blind test set, and will report their scores to the group.
Participants will learn through this hands-on data science activity:
Challenges in evaluating performance without a gold standard answer
The trade-off between sensitivity and positive predictive value in building a predictive model
The challenges in understanding and processing clinical text
The cost and benefit of a rule-based system
Participants will also demonstrate the following skills in relation to teaching data science:
Describe situations in which teaching data science at a high level without requiring programming may be useful
Explain how Jupyter Notebooks can be integrated in informatics education to support both novice and experienced students
Explain how Jupyter Notebooks can be used to explore data, experiment with and evaluate tools and algorithms, and visualize information
Wendy Chapman (Presenter)
University of Utah
Brian Chapman (Presenter)
University of Utah