Overview

Details

This session contains a 40 minute tutorial and a 40 minute workshop with a 10 minute break in between.

Much of the information documented in the Electronic Health Record (EHR) during clinical care is stored in free-text narratives, making them difficult to access on a large scale. Natural Language Processing (NLP) empowers researchers and developers to unlock that crucial information from clinical texts. Recent advances in NLP have made the technology more accessible to users with both technical and non-technical backgrounds, making NLP a powerful tool for anyone using EHR data.
Participants will have the opportunity to follow along and program a simple NLP pipeline. Code will use a web-based environment like Google Colab that allows participants to write or execute code and see results with no installation required. Step-by-step instructions for each phase of the workshop will be provided and code will remain available after tutorial completion.
This tutorial will provide an overview to clinical NLP and an applied introduction to using NLP to extract information from clinical text. With the open-source Python library medspaCy[1], participants will use rule-based methods, ontologies, and pre-trained machine learning models to gain hands-on experience with NLP by assembling, refining, and executing clinical NLP pipelines using synthetic clinical documents.

• Frame clinical problems in a way that can be addressed by NLP
• Understand common methods and algorithms used to solve an NLP problem
• Integrate both rule-based and machine learning methods into an NLP pipeline
• Iteratively develop an NLP pipeline using medspaCy
• Evaluate and discuss errors produced by an NLP pipeline

The goal of this workshop is to provide an overview of what are the necessary steps to be followed when applying process mining in the healthcare domain.
The aims of the workshop session are: (1) identify the main steps required to execute a process mining study using healthcare data, (2) overview some of the process mining tools available in field (research and industry), (3) identify the challenges when using clinical data for the identification, generation and analysis of events and activities for process mining, and (4) demonstrate the use of a case study with simulated data on how to accomplish some guided question analysis using process mining.

Speaker