-
https://d1y502jg6fpugt.cloudfront.net/13418/archive/files/74aa8f4039f5bcb2d69e3cfeba71adea.pdf?Expires=1712793600&Signature=oXxmThz9nLj6rvLHuOHPos6lU0p9zY-%7E98tZ2tnC62ZJ92DiXD2Yr62lMa2hBzD77JmAFeTgZHQoT3H2FvLdq8AYO4bWY7clAnPVOiyA%7EeZA6DKwxMb8yOK9kCPCoFYAx-7qSztujMQ9967JBKRFTMfkftbYD5cseDKtA9wSWuf%7EkJkZrcBIXFtAPrm5IOowu6DgD5CnVG1UmWzc1WepbtgbxwnGFlF6IQc%7E7eXq2hYuUc0scvwz3F3AMebxxEqJNNWhHsqS9cUZSucKYo%7EqLCj7Zi9-zQ1o6pl1epM10duWtjqt1iyomb9iR6cVzoGiv3TSICuCqJrChchMuFHq6g__&Key-Pair-Id=K6UGZS9ZTDSZM
5f7398771ec0a33341cc02b2613ede5c
PDF Text
Text
Data Mining with MIMIC‐II in BIDMC ICUs
Joseph Paonessa MD, Thomas Brennan PhD, Mengling Feng PhD, Roger Mark MD PhD, Leo Anthony Celi MD, MS, MPH
Beth Israel Deaconess Medical Center – Harvard Medical School
MIT Laboratory of Computational Physiology
Problem
Progress
The failure to store and analyze the vast amount of
data generated on a daily basis is a key hurdle in
advancing the practice of critical care medicine. The
intensive care unit (ICU) provides a cogent example of
a data rich clinical domain in which an insufficient
portion of the data generated has been employed for
guiding practice by, for example, supporting the
creation of clinical decision support tools, identifying
significant patterns in population data, and employing
feedback on system outputs for the formulation of
systematic process improvements.
Predictive Modeling, Prognostication, and Outcomes
The MIMIC database has allowed our group to develop predictive models with actionable
outputs that potentially lead to measurable improvements in process and/or outcome. Such
models could support appropriate early triage regarding level of care and monitoring, as well as
the allotment of costly resources such as specialist‐requiring interventions and/or technologies.
For example, these tools could assist emergency departments if limitations in ICU resources lead
to regionalization of critical care.
Aims
Unraveling Complexity and Variability
MIMIC that include detailed clinical information has provided researchers an opportunity to
accumulate safety and efficacy evidence, discover patient subpopulations that experience
important variances in efficacy or unanticipated delayed adverse effects, and uncover
interactions between and among simultaneous treatments as drugs become used in wider, more
diverse patient populations than those possible during premarket approval clinical studies.
To build a learning system around an open‐access ICU
database where practice informs research, and
research informs practice. Clinicians at the frontline of
care should be at the core of this dynamic learning
system, fully supported by engineers to collaborate on
the daily translation of questions into strategies for
database interrogation, modeling and analysis.
The Teams
Teams of clinicians (nurses, doctors, pharmacists) and
scientists
(database
engineers,
modelers,
epidemiologists) have formed around the Multi‐
parameter Intelligent Monitoring in Intensive Care
(MIMIC) database. The database was established in
October 2003 from a partnership that combines the
resources of a powerful interdisciplinary team from
academia (Massachusetts Institute of Technology),
industry (Philips Medical Systems and Philips
Research North America) and clinical medicine (Beth
Israel Deaconess Medical Center). The public‐access
database now holds over 60,000 ICU stays in the
BIDMC ICUs.
The inter‐disciplinary teams have been translating
day‐to‐day questions typically asked during rounds
that often have no clear answers in the current
medical literature into study designs and then
perform the modeling and the analysis and publish
their findings.
Intervention
The learning system described above has been
operating since 2010. The scientists attend ICU
rounds, observe the processes surrounding data
capture and interact with the rest of the clinical team.
The clinicians on the other hand come to MIT and
learn the tenets of clinical data analysis. This culture
of collaboration is crucial in democratizing research
and lowering the barrier for frontline clinicians who
are most familiar with the information gaps in
practice to participate in knowledge generation, as
well as for those not traditionally associated with
evidence creation, including patients themselves.
In January 2014, the MIT‐BIDMC Critical Data Marathon and Conference was held. The basic
premise of the data marathon was to bring together providers and data experts to answer
clinically‐relevant questions over the course of a weekend. While a truly novel discovery or a
fully functioning solution is a rare outcome, these events enable crowdsourcing of valuable,
varied points‐of‐view and new personal connections that will form the basis for longer‐term
collaborations. More than 80 participants formed 10 teams, including one in London, United
Kingdom. The best projects were presented during the Critical Data Conference that followed.
The conference was attended by more than 200 participants, and watched by another 400 via a
live stream. The talks and discussions revolved around two themes: how to operationalize the
vision of a data‐driven learning system and how to safeguard big data in healthcare from
contributing further to the swaths of unreliable research that plague medicine. The
presentations have been made available online and have been viewed by more than 700
unique visitors (http://criticaldata.mit.edu/events/conference/program.html).
Next steps
While the current MIMIC database is limited to one academic hospital in the United States, plans
are already in motion to extend the data to other hospitals, including institutions outside the
U.S. As the database expands quantitatively and qualitatively across diverse care environments,
the power and significance of any individual analysis will only increase over time. Furthermore,
such analyses can be easily repeated, modified, and strategically improved based on iterative
interpretation of prior findings.
Our vision is for the development of a care system consisting of “clinical informatics without
walls”, in which the creation of evidence and clinical decision support tools is initiated, updated,
honed, and enhanced by crowd sourcing. In this collaborative medical culture, knowledge
generation would become routine and fully integrated into the clinical workflow. This system
would employ individual data to benefit the care of populations and population data to benefit
the care of individuals.
Acknowledgements
NIBIB Grant 2R01‐001659
�
Dublin Core
The Dublin Core metadata element set is common to all Omeka records, including items, files, and collections. For more information see, http://dublincore.org/documents/dces/.
Title
A name given to the resource
Silverman Symposium
Description
An account of the resource
Each year the Silverman Symposium poster session offers BIDMC staff and affiliates the opportunity to share experiences and learn about efforts to improve Quality and Safety.
Date
A point or period of time associated with an event in the lifecycle of the resource
2021
Silverman Poster
Primary Contact
If you would like more information about this project, contact this person. Make email address clickable.
Joseph Paonessa (<a href="mailto:jpaoness@bidmc.harvard.edu">jpaoness@bidmc.harvard.edu</a>)
Department
Any departments listed on the poster or identified in the spreadsheet.
Internal Medicine
BIDMC Location
The BIDMC location where the poster team resides if identified in spreadsheet. If not identified, choose BIDMC.
BIDMC
Project Team
Joseph Paonessa<br />Thomas Brennan<br />Mengling Feng<br />Roger Mark<br />Leo Anthony Celi
Dublin Core
The Dublin Core metadata element set is common to all Omeka records, including items, files, and collections. For more information see, http://dublincore.org/documents/dces/.
Title
A name given to the resource
Data Mining with MIMIC-II in BIDMC ICUs
Date
A point or period of time associated with an event in the lifecycle of the resource
2014
Format
The file format, physical medium, or dimensions of the resource
pdf
Effectiveness