Partners
HealthCare Embraces the Democratization of AI
to
Accelerate Innovation in Medicine
The
MGH & BWH Center for Clinical Data Science wants to put AI in the toolbox
of every researcher
and
clinician
Boston,
MA April 08, 2019 – At the World Medical Innovation Forum 2019, Partners
HealthCare today
announced
a commitment to put artificial intelligence (AI) in the toolbox of all of its
researchers and
clinicians.
These efforts, being spearheaded by the MGH & BWH Center for Clinical Data
Science (CCDS), focus
on providing researchers and clinicians with access to data, GPU (graphics
processing unit) compute capacity
and supporting software required to develop their own AI algorithms to be
implemented in the clinical
environment.
CCDS
is planning broad roll-out to the Partners system over the next 12 months and
is already offering AI
capabilities
and support services. This announcement aligns with national and global trends
towards
democratization
of AI capabilities through increasingly collaborative efforts between academic
organizations,
technology companies and organizations like the American College of Radiology
(ACR)
that
work to establish standards for the field.
“Currently,
focused and siloed pockets of domain expertise in AI reside within specific
departments or labs at
several large academic medical centers, but making AI an enabling technology
across the field of
healthcare
has been a challenge for many facilities” said Keith Dreyer, DO, PhD, FACR,
FSIIM, Chief
Data
Science Officer, Partners HealthCare. “The truth is, you don’t have to be a
computer scientist or data scientist
to participate in the creation of AI – we are just starting to see increasing
availability of tools to enable
on-premises development of AI models by clinicians.”
The
trend is punctuated by the recent announcement of a free software platform
(AI-LAB) being released
by
the ACR to allow radiologists to develop, validate and implement AI within
their own facilities. AILAB
is
a core component of ACR’s vision for the democratization of AI in radiology.
The free, vendorneutral
software
solution will allow radiologists to work with patient data within their own
secure IT
environments,
while providing opportunities to share AI datasets and algorithms with other institutions.
The
objective is to facilitate the development of more versatile, generalizable and
accurate models – a
principle
known as transfer learning.
Through
collaboration with Nvidia and The Ohio State University (OSU), CCDS has
demonstrated the
ability
for new transfer learning approaches to generate algorithms that are more
robust and adaptive to
local
variation. The specific project involved the transfer of a cardiac Computed
Tomography Angiography
(CTA)
model, which was co-developed with Nvidia, to OSU where it was re-trained using
OSU data. The
resulting
model was more accurate and required less effort to train, validate and test.
Ittai Dayan, MD,
Executive
Director of Operations at CCDS underscores the spirit of collaboration. “The
concept of transfer learning
means that sites can collaborate in building more highly performant algorithms
without having to share
sensitive patient data – this collaborative approach will dramatically
accelerate the rate at which new algorithms
are developed and integrated into the clinical workflow.”
The
announcement at the World Medical Innovation Forum in Boston, a global
conference focusing on the
intersection
of AI and clinical care, reflects the vital role of innovation in transforming
healthcare. The
ability
for new technologies, like AI, to have a meaningful impact in the field of
medicine is highly
dependent
on collaboration between stakeholders with a unified goal of improving patient
lives. At the
conference,
clinicians, industry and government leaders, entrepreneurs and venture
investors are exploring
where
AI meets clinical care, including the requisite sensitivity to patient data and
standards.
To
see more of CCDS’ work at the Forum, visit the CCDS booth onsite on the 3rd floor
of the Westin
Copley.
沒有留言:
發佈留言