Embed Size (px)
Citation preview
30 November 2017 1
DIGITAL LEARNING COLLABORATIVE of the
NAM Leadership Consortium for a Value & Science-Driven Health System
Presented by: Jim Fackler, MD
Artificial Intelligence and the future of continuous health learning and improvement
Disclosures 1
• Machine Learning for Healthcare – Board Member - www.mlforhc.org – August 17-18th at Stanford
• Papers due 20 April 2018 • Published in JMLR (about 30% acceptance
rate)
2 30 November 2017
Disclosures 2
• Rubicon Health – Founder – Startup to offer Team Performance
Optimization with Augmented Intelligence
• Under a licensing agreement between Rubicon Health and the Johns Hopkins University, Dr. Fackler is entitled to royalty distributions on technology described in this presentation. Dr. Fackler is also the founder of and holds equity in Rubicon Health.
• This arrangement has been reviewed and approved by the Johns Hopkins University in accordance with its conflict of interest policies.
3 30 November 2017
30 November 2017 4
Artificial Intelligence and the future of continuous health learning and improvement
Presented by: Jim Fackler, MD
This session will focus on the role of data integration and sharing in enhancing the capabilities of machine learning algorithms to improve health and health care.
The role of data integration
5 30 November 2017
The role of data integration
6
6,853 ft2 -- clinical 780 ft2 -- storage
The role of data integration
7
6,853 ft2 -- clinical 780 ft2 -- storage
30 November 2017 8
Artificial Intelligence and the future of continuous health learning and improvement
Presented by: Jim Fackler, MD
This session will focus on the role of data integration and sharing …
Point 1 – there are 350 data streams per patient Point 2 – 100’s of care team members need AI
The role of data integration
9 30 November 2017
© Springer International Publishing Switzerland 2016 C.A. Weaver et al. (eds.), Healthcare Information Management Systems: Cases, Strategies, and Solutions, Health Informatics.
Pre-existing Patient conditions data Location Genome portal Remote Environment wearable EMS EMR Exogenous implanted ED Devices Endogenous Hospital Diagnostic Chronic disease ICU Therapeutic Physiological monitor Proximate Issues Location Exposures Home Surgery Nursing home Trauma LTAC SNIF
The role of data integration
10 30 November 2017
30 November 2017 11
Artificial Intelligence and the future of continuous health learning and improvement
Presented by: Jim Fackler, MD
This session will focus on the role of data integration and sharing …
Point 1 – there are 350 data streams per patient Point 2 – 100’s of care team members need AI Point 3 – SMART with FHIR / federated data stores
Pre-existing Patient conditions data Location Genome portal Remote Environment wearable EMS Exogenous implanted ED Endogenous Hospital Chronic disease ICU Proximate Issues Location Exposures Home Surgery Assisted Living Trauma LTAC SNIF
12 30 November 2017
Sepsis – Roles for AI from Health through Sepsis to Health
Biological Symptoms Signs Physiology Diagnostic Lab Therapeutic Interventions Outcomes background decisions decisions
T0=Bacterial Invasion
TC-D TBC-O TAI-D TBC Tabx-O Health
Morbidity Death
Tlact
Tvaso-O
Tetc-O
Tfluid-O
Tabx1
Tfluid1
Tvaso1
Tetc1
Pre-existing Patient conditions data Location Genome portal Remote Environment wearable EMS Exogenous implanted ED Endogenous Hospital Chronic disease ICU Proximate Issues Location Exposures Home Surgery Assisted Living Trauma LTAC SNIF
13 30 November 2017
Sepsis – Roles for AI from Health through Sepsis to Health
Biological Symptoms Signs Physiology Diagnostic Lab Therapeutic Interventions Outcomes background decisions decisions
T0=Bacterial Invasion
TC-D TBC-O TAI-D TBC-RSLT Tabx-O Health
Morbidity Death
Tlact
Tvaso-O
Tetc-O
Tfluid-O
Tabx1
Tfluid1
Tvaso1
Tetc1
Iteratively
FIGURE 3-1 Outcomes from the diagnostic process. NAM 2015
The value of data integration
14 30 November 2017
A cPOP is a complete
Performance Optimization
Package
Trigger Physiological /
EMR data alerting for sepsis
Think Focus clinician awareness of data and aid decisions to
screen and test for sepsis
Treat Efficient delivery
of correct therapies to
septic patients
Track Iterate with
physiological / EMR data to implement
treatments and monitor patient responses to
therapies.
Diagnostic Stewardship
Therapeutic Stewardship
Workflow Optimization
30 November 2017 15
Artificial Intelligence and the future of continuous health learning and improvement
Presented by: Jim Fackler, MD
This session will focus on the role of data integration and sharing …
Point 1 – there are 350 data streams per patient Point 2 – 100’s of care team members need AI Point 3 – SMART with FHIR / federated data stores Point 4 – AI will be useful when we continuously harvest the “data exhaust” to optimize performance and track outcomes.
30 November 2017 16
Artificial Intelligence and the future of continuous health learning and improvement
Presented by: Jim Fackler, MD
This session focused on the role of data integration and sharing …
Thank you [email protected] @jimfackler 443-858-6370
