Events Add an event Speakers Talks Collections
 
Duration 03:05
16+
Video

A9 Knowledge-Base Completion for Constructing Problem-Oriented Medical Records

James Mullenbach
Research Engineer at ASAPP
+ 4 speakers
  • Video
  • Table of contents
  • Video
Machine Learning for Healthcare
August 7, 2020, Online, Los Angeles, CA, USA
Machine Learning for Healthcare
Request Q&A
Machine Learning for Healthcare
From the conference
Machine Learning for Healthcare
Request Q&A
Video
A9 Knowledge-Base Completion for Constructing Problem-Oriented Medical Records
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Add to favorites
533
I like 0
I dislike 0
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
  • Description
  • Transcript
  • Discussion

About the talk

00:00 Intro

00:15 Background

00:40 Atria fibrillation use cases

01:26 Technical challenge

01:53 Constructing the initial graph

02:03 DistMult model

02:18 Experimental results

About speakers

James Mullenbach
Research Engineer at ASAPP
Jordan Swartz
Associate Professor / Physician Informaticist at NYU Langone Health
Greg McKelvey
Assistant Director for Biosecurity at White House Office of Science and Technology Policy
Hui Dai
VP of Machine Learning at ASAPP
David Sontag
PhD, Associate Professor of Electrical Engineering and Computer Science at MIT

I'm guessing: "I am a rising senior in Computer Engineering at the ASAPP. In my free time, I enjoy playing basketball, watching football, cooking, and trying new food restaurants. My main areas of interest include machine learning, neural networks, data analytics, software development, robotics, computer vision, artificial intelligence, web development and testing, game design, graphic design, animation, video editing, art/design, music production, etc"

View the profile

I divide my time between caring for patients, teaching residents, and using IT to improve patient care.

View the profile

Dr. T. Greg McKelvey Jr. MD, MPH, is VP of Health at ASAPP. In his role, Greg leads teams of doctors, designers, data scientists and software developers to create AI-native products that enhance patient engagement, access, and experience through augmented interactions. Prior to his role at ASAPP, Greg was the Chief Medical Officer at KenSci. He trained in Plastic & Reconstructive Surgery at Montefiore Medical Center,in Occupational & Environmental Medicine and Biomedical & Health Informatics at the University of Washington (UW). He has also co-authored several publications like Development and Prospective Validation of a Machine Learning-Based Risk of Readmission Model in a Large Military Hospital, Predicting Likelihood of Emergency Department Admission Prior to Triage: Utilising Machine Learning within a COPD Cohort, Predicting Future Frequent Users of Emergency Departments in California State, to name a few. Greg received his Master of Public Health (M.P.H) from the John Hopkins Bloomberg School of Public Health and his Doctor of Medicine (MD) from Dartmouth Medical School.

View the profile

David Sontag joined the MIT faculty in 2017 as Hermann L. F. von Helmholtz Career Development Professor in the Institute for Medical Engineering and Science (IMES) and as Associate Professor in the Department of Electrical Engineering and Computer Science (EECS). He is also a principal investigator in the Computer Science and Artificial Intelligence Laboratory (CSAIL). Professor Sontag’s research interests are in machine learning and artificial intelligence. As part of IMES, he leads a research group that aims to transform healthcare through the use of machine learning. Prior to joining MIT, Dr. Sontag was an Assistant Professor in Computer Science and Data Science at New York University’s Courant Institute of Mathematical Sciences from 2011 to 2016, and postdoctoral researcher at Microsoft Research New England from 2010 to 2011. Dr. Sontag received the Sprowls award for outstanding doctoral thesis in Computer Science at MIT in 2010, best paper awards at the conferences Empirical Methods in Natural Language Processing (EMNLP), Uncertainty in Artificial Intelligence (UAI), and Neural Information Processing Systems (NIPS), faculty awards from Google, Facebook, and Adobe, and a NSF CAREER Award. Dr. Sontag received a B.A. from the University of California, Berkeley.

View the profile
Share

Hi. I'm dangling back and I'm going to prison or work on knowledge base completion for constructing problem. Oriented medical records, come with my colleagues at ASAP. There's the problem in healthcare that was once called Death by a thousand klicks. We're in a position, often spends more time interacting with their EHR software. To find patient data relevant to their decision-making in a source-oriented records when did it could instead be oriented around health problems. As first suggested, half, a century ago, by Lawrence weed oriented, by example

in treating a patient with atrial fibrillation and irregular heartbeat and determine if the dosage of their blood, thinner is appropriate and a technical source-oriented record of today. As in this mock-up, you have to navigate the different sources of data for this many on the left to find medications. Then you have to dig through their medication history, to find the dosage, make a mental note of it, go back to the menu, click on our results, look at the lab results and find their clotting parameters. And then keep the dosage in mind to make a decision. So instead of all that

clicking and searching instead oriented data around, the problem itself of April for relation and show the medications procedures and lab results, all underneath the problem headache, and we can do this for all the Patients health problems. Do the technical challenge. We have to determine how to actually associate the relevant medication procedures and laughs, with the problems we could use relations in your MLS, or other ontology is, but those are always incomplete. Bearings my baked these relationships in and we find that are useful but they're not enough. So we take the approach of

treating, this data is knowledge graph or knowledge base to be completed, which we know we can train machine learning models to do. To construct the starting graph. We defined 32 problems that are common in our data. Set to take several dozen medication Labs the procedures for each of those. Between adjustment model, which was previously introduced for general knowledge base completion, which competes a three-way top product of problem, relation in Target, in Billings and return this with their annotations, which hopefully includes examples. Experiments. We can pair using embeddings

trained internally to a site site specific embeddings compared to the external open-source embeddings. Trained on larger data and compare on the main receptacle Rank and hits at 5 Frankie and the external open-source, the Medics turn out a little bit better, just using them out of the box when we use them to initialize with, this won't model and trained that we get a bit of a line using engineered features. On top of that, I got provides a boost your some example outputs for the best-performing model on the meaning of sleep apnea. And there's a separate set of results in the paper that

show that we can use Canon to improve initialization. Thanks. But that all of our code and data is open-sourced happy to take questions at the poster session.

Cackle comments for the website

Buy this talk

Access to the talk “A9 Knowledge-Base Completion for Constructing Problem-Oriented Medical Records”
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free

Ticket

Get access to all videos “Machine Learning for Healthcare”
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Ticket

Interested in topic “Artificial Intelligence and Machine Learning”?

You might be interested in videos from this event

February 4 - 5, 2021
Online
26
115
ai, application, bot, chatbot, conversation, data, design, healthcare, ml

Similar talks

Ariane Marelli
Professor of Medicine at McGill University
+ 10 speakers
Chao Chao Li
E. B. Eddy Professor and Canada Research Chair at McGill University
+ 10 speakers
Aihua Liu
Psychologist at McGill University
+ 10 speakers
Hanh Nguyen
Рsychologist at McGill University
+ 10 speakers
James Brophy
Professor at McGill University
+ 10 speakers
Liming Guo
Research at McGill University
+ 10 speakers
David Buckeridge
Professor at McGill University
+ 10 speakers
Jian Tang
Assistant Professor at HEC Montreal
+ 10 speakers
Joelle Pineau
Associate Professor at McGill University
+ 10 speakers
Yi Yang
Associate Professor of Statistics at McGill University
+ 10 speakers
Yue Li
Assistant Professor at McGill University
+ 10 speakers
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Stephanie Skove
Bachelor of Science at Duke University
+ 12 speakers
Harvey Shi
Associate in Research at Duke Institute for Health Innovation (DIHI)
+ 12 speakers
Ziyuan Shen
Software Engineer at ZipRecruiter
+ 12 speakers
Michael Gao
Data Science Lead at Duke Institute for Health Innovation
+ 12 speakers
Mengxuan Cui
Statistical Modeling Engineer at Yidu Tech Inc.
+ 12 speakers
Marshall Nichols
Senior IT & Analytics Manager at Duke Institute for Health Innovation
+ 12 speakers
Suresh Balu
Program Director at Duke School of Medicine
+ 12 speakers
Cara O’Brien
Assistant Professor at Medicine at Duke University
+ 12 speakers
Armando Bedoya
Associate Chief Medical Informatics Officer, Duke University Health System at Duke Health Technology Solutions
+ 12 speakers
Dustin Tart
Duke University at Program Manager, Patient Response
+ 12 speakers
Benjamin Goldstein
Associate Professor at Duke University
+ 12 speakers
Mark Sendak
Population Health & Data Science Lead at Duke Institute for Health Innovation
+ 12 speakers
William Ratliff
Innovation Program Manager at Duke University Health System
+ 12 speakers
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Emma Brunskill
Assistant professor of computer science at Stanford University
+ 1 speaker
Finale Doshi-Velez
Gordon McKay Professor in Computer Science at Harvard University
+ 1 speaker
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free

Buy this video

Video
Access to the talk “A9 Knowledge-Base Completion for Constructing Problem-Oriented Medical Records”
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free

Conference Cast

With ConferenceCast.tv, you get access to our library of the world's best conference talks.

Conference Cast
947 conferences
37617 speakers
14378 hours of content
James Mullenbach
Jordan Swartz
Greg McKelvey
Hui Dai
David Sontag