Past Meetings

Uses of Natural Language Processing in HealthcareBack

03/17/2017

Our March 17th Health TechNet meeting focused on the new and developing uses of Natural Language processing in several key areas of healthcare data processing in the research, care delivery, and insurance communities.
 
What is Natural Language Processing? Natural language processing (NLP) might be defined as the ability of a computer to understand human speech as it is spoken and free text as it is written.  The use of this technique for the gathering and entry of data, and its subsequent processing and usefulness, is of growing significance in healthcare.  Since language is often ambiguous, incomplete and fractal, machine learning techniques are often paired with NLP, as will be discussed.  At our meeting, we’ll have two leading expert speakers knowledgeable on this subject.  Our moderator, Joe Bormel first established the context within the Health TechNet community's policy and technology educational series.

 
Joseph Polifroni, Ph.D. is the Director, Natural Language Processing for Optum Analytics who gave a presentation titled “Promise and Reality:  Tales from the Intersection of Big Data, Machine Learning, and Healthcare Analytics”.  His presentation focused on the importance of understanding the potential and the limitations of medical notes data, the importance of phrasing questions in actionable ways, and the importance of early examination of the data themselves. Dr. Polifroni has been involved in the fields of Automatic Speech Recognition and Natural Language Processing for over 25 years.  He began his career as a Research Scientist in the Spoken Language Systems Group at MIT’s Laboratory for Computer Science.  He has worked on machine learning for the design and evaluation of spoken dialogue systems, sentiment analysis, named entity extraction, and user modeling in both industry and academe.  He also worked on the speech/natural language interface to a home healthcare monitoring system for the elderly deployed by a hospital group in Taiwan.
 
Guy Divita, MS a computer scientist, author, standards developer and industry leader gave a short version of his presentation entitled "The Magic Behind Clinical Natural Language Processing for Epidemiological Studies, Surveillance and Quality Metrics”.  (Due to time constraints, he is scheduled to complete this presentation at our next meeting on April 21.)  He is currently a Research Associate at University of Utah Medical School in the Department of Bioinformatics.  Divita has lead and published multiple large scale projects including recent work for VA Salt Lake City Health Care System (IDEAS Center) which will be shared at our meeting.  Divita has worked in academia, industry and government at the National Library of Medicine, in the field of Big Data and NLP for several decades and is broadly recognized as a leading expert and platform creator/developer.  
 
Joseph Bormel, MD, is a board certified internist, and NLM fellowship-trained informatics professional.  He has several decades of executive experience with publicly traded EHR vendors (with Cerner, QuadraMed and the EHRA), program officer experience  (with HHS, Office of the National Coordinator for HIT), and start-up experience (with San Francisco-based Healthline Networks and Talix).  Bormel shared several applied case studies, elaborating contracted project work done applying Natural Language Processing and controlled vocabularies groupings to value-based purchasing, in both the public and private sectors.  Specifically, he described the application of NLP to improving documentation of care needed for Medicare Advantage programs to address risk adjustment.  He also described the relationship of NLP to quality measures that are used in the MACRA program, called electronic clinical quality measures, based on his measures-authoring work for Medicare.  Lastly, he shared some recent research on how several government agencies are evolving their bio surveillance to include EHR+NLP derived data to transcend the well-known limits of claims data.           

Dr. Bormel’s presentation is attached and Mr. Polifroni’s is available upon request, via email at david.main@nelsonmullins.com.  Mr. Divita’s presentation will be posted in the summary of our April, 2017 meeting.