professor of medicine, physiology, and biomedical engineering, … · 2015. 4. 14. · blee...

40
J. Randall Moorman, M.D. Professor of Medicine, Physiology, and Biomedical Engineering, University of Virginia

Upload: others

Post on 04-Feb-2021

1 views

Category:

Documents


0 download

TRANSCRIPT

  • J. Randall Moorman, M.D.Professor of Medicine, Physiology, and Biomedical Engineering,

    University of Virginia

  • Predictive monitoring for early detection of subacute, potentially catastrophic

    illnesses

    On behalf of many:J Randall Moorman, MD

    Professor of Medicine, Physiology, BMEUniversity of Virginia

    CMO, Adult Medical Predictive Devices, Diagnostics, DisplaysShareholder, Medical Predictive Science Corporation

    Vice-president, Society for Complex Acute IllnessEditor-in-chief, Physiological Measurement

  • Predictive [email protected]

    Randall Moorman, M.D.*

    Doug Lake, Ph.D.*

    Karen Fairchild, M.D. John Kattwinkel, M.D.

    Forrest Calland, M.D.

    Travis Moss, M.D. Kyle Enfield, M.D.

    Robert Sinkin, M.D.

    George Stukenborg, Ph.D. Chris Moore, M.D.

    Matthew Clark, Ph.D.*

    Laura Barnes, Ph.D. John Delos, Ph.D.

    Christian Rieser, Ph.D.

    Milan, Columbia, MGH, Emory, UCSF, MITRE …

    Mary Mohr (Ph.D. 2015)

    Marta Carrara, M.S.

    Luca Carozzi, M.S.

    Marco DePasquale (M.S. 2015)

    Nate Ivanek, M.D.

    Diana Gomez, M.D.

    Eric Holland, M.D.

    Rich Kronfol, M.D.

    Brynne Sullivan, M.D.

    Manisha Patel, M.D.

    Christina McClure (M.D. 2015)

    Abel David (M.D. 2018)

    Caroline Ruminski (M.D. 2018)

    Blee Moffett, RN

    Cathy Horton, RN

    *Equity, consulting, employment: MPSC, AMP3D, Cville

  • Mission

    • Save lives

    • Multicenter development and validation of predictive algorithms

    Precepts

    • Apparently sudden critical illness can have sub-clinical prodromes with illness signatures

    • Diagnosis and treatment at this early stage can save lives

  • Is this baby septic?We developed a predictive monitor for early detection

  • Who is sick in my NICU?

  • Mortality reduced 10.2% 8.1% with display of predictive monitor

    One extra survivor per 48 VLBW, 23 ELBW, and

    12 septic infants monitored

    Predictive monitoring

    Conventional monitoring

  • Predictive monitoring in adults

    • 3 subacute potentially catastrophic illnesses:• Sepsis (>3X ICU stay, >3X mortality)

    • Bleeding leading to large transfusion (>3X, >3X)

    • Lung failure leading to urgent unplanned intubation (>4X, >4X)

    • 2 ICUs: Medical and Surgical/Trauma/Burn

    • >7000 patients with >1000 events

    • >70TB data

    • Method: Big Data, high performance computing

    • Result: efficient algorithms deployed by low tech

  • Sepsis Bleeding Lungfailure

    Death

    Big Data reveals illness signatures

  • Sepsis Bleeding Lungfailure

    Death

    UnitROC AUC# events

    Big Data reveals illness signatures

  • Who’s sick in my Surgery ICU?

    BP

    T

    HRO2

    RR

  • STBICU3/12 07:00

  • STBICU3/12 08:00

  • STBICU3/12 09:00

  • STBICU3/12 10:00

  • STBICU3/12 11:00

  • STBICU3/12 12:00

  • STBICU3/12 13:00

  • STBICU3/12 14:00

  • STBICU3/12 15:00

  • STBICU3/12 16:00

  • STBICU3/12 17:00

  • STBICU3/12 18:00

  • STBICU3/12 19:00

  • STBICU3/12 20:00

  • STBICU3/12 21:00

  • STBICU3/12 22:00

  • STBICU3/12 23:00

  • STBICU3/13 00:00

  • STBICU3/13 01:00

  • STBICU3/13 02:00

  • STBICU3/13 03:00

  • STBICU3/13 04:00

  • STBICU3/13 05:00

  • STBICU3/13 06:00

  • STBICU3/13 07:00

  • STBICU3/13 08:00

  • STBICU3/13 09:00

  • Relevance to healthcare in the US?

    We envision a stable framework for a national multicenter

    collaborative effort to develop predictive monitoring and other

    clinical analytics tools we can all use to save lives with.