- Methods and Applications of Natural Language Processing in Medicine
- Large Scale Ensembled NLP Systems with Docker and Kubernetes
- The Overview Effect: Clinical Medicine and Healthcare Concepts for the Data Scientist
Rui Zhang, University of Minnesota, USA
Yanshan Wang, Mayo Clinic, USA
Hua Xu, University of Texas Health School of Biomedical Informatics, USA
Yifan Peng, National Center for Biotechnology Information, USA
Tuesday August 25, 09:00-12:00 CDT
Rapid growth in adoption of electronic health records (EHRs) has led to an unprecedented expansion in the availability of large longitudinal datasets. Large initiatives such as the Electronic Medical Records and Genomics (eMERGE) Network, the Patient-Centered Outcomes Research Network (PCORNet), and the Observational Health Data Science and Informatics (OHDSI) consortium, have been established and have reported successful applications of secondary use of EHRs in clinical research and practice. In these applications, natural language processing (NLP) technologies have played a crucial role as much of detailed patient information in EHRs is embedded in narrative clinical documents. Meanwhile, a number of clinical NLP systems, such as MedLEE, MetaMap/MetaMap Lite, cTAKES, MedTagger, and i2b2/n2c2 have been developed and utilized to extract useful information from diverse types of clinical text, such as clinical notes, radiology reports, and pathology reports. Success stories in applying these tools have been reported widely.
NLP is considered as a branch of AI. Despite the demonstrated success of NLP in many applications, methodologies and tools developed for the clinical NLP are still underknown and underutilized by experts and physicians in the clinical domain. In this tutorial, we will introduce NLP methodology and their applications in medicine. We group a panel of NLP researchers from four institutions: University of Minnesota, Mayo Clinic, US National Library of Medicine and University of Texas Health Science Center at Houston.
Through this half-day tutorial, we would like to introduce our methodological efforts in applying NLP to the clinical domain, and showcase our real-world NLP applications in clinical practice and research across four institutions. We will review NLP techniques in solving clinical problems and facilitating clinical research, the state-of-the art clinical NLP tools, and share collaboration experience with clinicians, as well as publicly available EHR data and medical resources, and finally conclude the tutorial with vast opportunities and challenges of clinical NLP. The tutorial will provide an overview of clinical backgrounds, and does not presume knowledge in medicine or health care.
Raymond Finzel, University of Minnesota, USA
Greg Silverman, University of Minnesota, USA
Shreya Datar, University of Minnesota, USA
Sijia Liu, Mayo Clinic, USA
Tuesday August 25, 13:00-16:00 CDT
Natural Language Processing (NLP) is a broad topic of interest for the Artificial Intelligence in Medicine community, and Docker and Kubernetes remain both buzzwords and useful technologies for industry and academia. This half-day tutorial aims to highlight ensemble methods within NLP (from basic to more cutting edge) while providing a general intro to Docker and Kubernetes. We hope that participants will leave the session feeling empowered to develop their own NLP pipelines with Docker and Kubernetes in mind.
This half-day tutorial will provide attendees with the core skills and working knowledge necessary to implement complex NLP workflows using cloud native technologies. The tutorial will give practical introductions to UIMA-AS (Unstructured Information Management Architecture - Asynchronous Scaleout), Docker (a container protocol used for packaging applications), Kubernetes (a platform system for orchestrating the creation and deployment of containers), and Argo (a Kubernetes workflow manager). The tutorial will start with a basic overview and demonstration on how to compose NLP system ensembles from publicly available systems like cTAKES, MetaMap, CLAMP, and BioMedICUS. We will specifically show how to design NLP ensembles that have been optimized for particular clinical domains.
We will then proceed through an introduction to emerging container and cloud technologies–including providing attendees with the core concepts and technical terms required to understand the various technologies being utilized. Explanations of how, when, where, and why to use each technology, along with some of the practical challenges of using each in a high-security (PHI compliant) environment will be discussed. In the final portion, tutorial participants will participate in a hands-on session by using Docker and Kubernetes from their own computers (as installed on cloud servers provided by the tutorial organizers) and running a test NLP workflow for acronym disambiguation on a collection of clinical note excerpts (from the CASI corpus).
Anthony C. Chang, Children’s Hospital of Orange County, USA
David Ledbetter, Children’s Hospital of Los Angeles, USA
Dennis Wall, Stanford School of Medicine, USA
Tuesday August 25, 13:00-16:00 CDT
The overview effect that astronauts described during spaceflight while viewing Earth from outer space is a profound cognitive shift in awareness of our collective fragile existence. This unique perspective of looking back at Earth, perhaps even greater as you travel further in space, has great implications for medicine and health care from a data science and artificial intelligence viewpoint. For the data scientists in health care, it would be of paramount importance to have a similar overview effect by having a deeper understanding of clinical medicine and healthcare. These data scientists will become knowledgeable about how clinicians think as well as the nuances of health care system and data. For this group, clinical venues such as outpatient clinics and rounds at hospitals will serve as clinical training grounds for data scientists and scenarios will be presented for group discussion. Lastly, there can be discussions on the real problems and relevant issues in clinical medicine and healthcare to avoid the “perfect model but little impact” problem with some data science projects in health care.
This half-day tutorial aims to improve the understanding and appreciation of both clinical medicine and data science for both the clinician and the data scientist. This interface between clinical medicine and data science is more important than ever before as these two domains converge this decade. Often these two domains are in symbiosis but not in synergy as there are a myriad of nuances in each of these areas that lead to misconceptions. Topics covered include: how doctors think (biases and heuristics, system 1 vs 2 thinking, etc.), evidence-based medicine (vs intelligence-based medicine), the healthcare system and medical record, the healthcare data conundrum, and the issues with data science in medicine (dichotomy and automation bias, accuracy vs impact, etc.).
Martin Michalowski, University of Minnesota, USA
Robert Moskovitch, Ben-Gurion University, Israel
Wednesday August 26, 09:00-13:00 CDT
The human race is facing in the past months one of the most meaningful public health emergency in the modern era caused by the COVID-19 pandemic. This pandemic has brought various challenges, which resulted in lock-downs with significant economical costs and limiting life the way we knew it before. In this pandemic, some patients may be infected and get severe symptoms, while other are being infected without symptoms, while others do not get infected, which makes managing such pandemic very challenging. However, given the available digital data in the modern era, artificial intelligence and generally data science can be a meaningful tool in coping with the various challenges rising by such an unexpected pandemic, such as outbreak prediction, risk modeling (who will get infected, and develop symptoms), testing strategy optimization, drug development, treatment repurposing, vaccine development, and more. In this workshop we will hear from experts, both clinicians and informaticians, on the challenges and problems data science and AI can address related to the global pandemic, and relevant deployments and experiences in gearing AI to cope with COVID-19.