Validating a Topic Model that Predicts Pancreatic Cancer from Latent Structures in the Electronic Medical Record

Project Summary

Furthering the work of a 2016 Data+ team in predictive modeling of pancreatic cancer from electronic medical record (EMR) data, students Siwei Zhang (Masters Biostatistics) and Jake Ukleja (Computer Science) spent ten weeks building a model to predict pancreatic cancer from Electronic Medical Records (EMR) data. They worked with nine years worth of EMR data, including ICD9 diagnostic codes, that contained records from over 200,000 patients.

Themes and Categories
Contact
Paul Benich
bendich@math.duke.edu

Project Results: The team began with exploratory data analysis that illustrated median times of appearance and frequency of specific ICD9 codes, with an eye toward understanding the relation between these statistics and pancreatic cancer diagnosis. They then trained a topic model which predicted past pancreatic cancer diagnosis with high accuracy (93 percent AUC) from ICD9 codes. Finally, they used the topic model outcomes to identify a pool of high-risk patients for potential future study.

Click here for the Executive Summary

Project Leads:

Lisa Satterwhite, PhD

James Abbruzzese, MD

Joseph Lucas, PhD

Project Manager: Tyler Massaro

Related People

Related Projects

United Nations Sustainable Development Goal 7 calls for universal access to affordable, reliable, sustainable, and modern energy. Researchers and practitioners around the world have responded to this call by producing a wealth of energy access data. While many data gaps still exist, are we capturing the fullest potential from the information and research we do have, and what it tells us about how to accelerate energy access? Power for All’s Platform for Energy Access Knowledge (PEAK) is an interactive knowledge platform designed to automatically curate, organize, and streamline large, growing bodies of data into digestible, sharable, and useable knowledge through automated data capture, indexing, and visualization. A team of students led by Rebekah Shirley will consult with Power for All to creatively visualize PEAK’s library, and to explore machine learning and natural language processing tools that can enable auto-extraction and visualization of data for more effective science communication.

Are there relative value opportunities in the global corporate bond markets?  
A team of students will work with Professor Emma Rasiel to understand whether an analysis of credit spreads on bonds issued by international firms in multiple countries over time can shed light on potential arbitrage opportunities. The team will have frequent opportunities to interact with analytics professionals at a leading financial advisory and asset management firm.

 

A team of students will consult with a leading financial advisory and asset management firm that is seeking to understand how big data can shed light on the secondary market for construction machinery. Students will explore a combination of publicly-available datasets that describe the used-machinery market and its potential implications as an indicator for the business cycle. There will be frequent interactions with analytical professionals from the firm.