Closing Plenary: Bill Stead

download Closing Plenary: Bill Stead

of 24

Transcript of Closing Plenary: Bill Stead

  • 8/9/2019 Closing Plenary: Bill Stead

    1/24

    Recalibrating InformaticsTrue North

    William W. Stead, M.D.Associate Vice Chancellor for Health Affairs

    Chief Strategy & Information OfficerMcKesson Foundation Professor of Biomedical

    Informatics and Medicine

    Disclosures: Co-inventor of two patient medical record products one licensed toMcKesson, Inc., and one licensed to Informatics Corporation of America (ICA) fromwhich I receive royalties through Vanderbilt University. Director of HealthStream, a publiccompany, compensated by meeting fees & an annual option grant.

  • 8/9/2019 Closing Plenary: Bill Stead

    2/24

    Capture & re-usestandardized data

    StandardizedData

    UnstructuredData

    Old New

    Idea Take 1

  • 8/9/2019 Closing Plenary: Bill Stead

    3/24

    Idea take 1 The Need for a Shift

    Gap between todays HCIT & what we need Root cause

    The Opportunity The idea take 2 Shifting EHR computational paradigm

    The Idea take 3

    Action steps

    Road Map

  • 8/9/2019 Closing Plenary: Bill Stead

    4/24

    Central Conclusions of NRC Report

    Current efforts aimed at nationwidedeployment of HCIT will not besufficient to achieve the vision of 21st

    century health care, and may evenset back the cause

    Success will require emphasis onproviding cognitive support(assistance for thinking about andsolving problems).

    In the near term, embracemeasurable health care qualityimprovement as the driving rationalefor HCIT adoption efforts.2009

  • 8/9/2019 Closing Plenary: Bill Stead

    5/24

    Information-Intensive Aspects of theIOMs Vision for 21st Century Health Care

    Comprehensive data on patients conditions, treatments &outcomes

    Cognitive support for health care professionals & patients tohelp integrate

    patient-specific data

    evidence-based practice guidelines & research results Tools to manage a portfolio of patients & to highlight problems

    as they arise

    Rapid integration of new instrumentation, biologicalknowledge, treatment modalities, and so on into a learninghealth care system

    Accommodation of growing heterogeneity of locales forprovision of care

    Empowerment of patients and their families in effectivemanagement of health care decisions and their implementation

    Stead WW, Lin HS. 2009. Computational technology for effective health care: immediate steps and strategicdirections. Comput Sci and Telecom Board, Nat Res Council. Washington: National Academies Press.

  • 8/9/2019 Closing Plenary: Bill Stead

    6/24

    University of PittsburghMedical CenterPittsburgh, PA

    Veterans Administration

    Washington, DC

    HCA TriStarNashville, TN

    Vanderbilt UniversityMedical CenterNashville, TN

    Partners HealthcareBoston, MA

    Intermountain Health CareSalt Lake City, UT

    University of California,San FranciscoSan Francisco, CA

    Palo Alto Medical FoundationPalo Alto, CA

    Site Visits

  • 8/9/2019 Closing Plenary: Bill Stead

    7/24

    Site Visit Observations

    Patient records are fragmented.

    Clinical user interfaces mimic paper without human factors

    & safety design. Biomedical devices are poorly integrated.

    Systems are used often to document what has been done,after the fact, for regulatory and legal uses.

    Support for evidence-based medicine and computer-basedadvice is rare.

    Clinical research activities are not well integrated intoclinical care.

    Legacy systems are predominant.

    Centralization is the predominant method ofstandardization.

    Implementations timelines are long and course changes areexpensive.

    Response times are variable and long down times occur.

  • 8/9/2019 Closing Plenary: Bill Stead

    8/24

    Data Mining

    Automation

    Connectivity DecisionSupport

    Root Cause: Mismatch between ComputationalTechnique & Scale of Problem

    Stead WW. Electronic Health Records. In: Rouse WB, Cortese DA, eds. Engineering the system of healthcare delivery.Tennenbaum Institute Series on Enterprise Systems, Vol. 3. Amsterdam: IOS Press; 2009.

  • 8/9/2019 Closing Plenary: Bill Stead

    9/24

    PATIENT CARE RESEARCH

    DecisionS

    upport

    Medical

    Logic

    VirtualPatient

    MedicalKnowledge

    Whereclinicianswant to stay

    Transactions

    Raw data

    Clinicalresearch

    transactions

    Raw researchdata

    WhereHealth ITchains us

    Workflow modeling and support, usability, cognitive support,computer-supported cooperative work (CSCW), etc.

    Stead WW, Lin HS. 2009. Computational technology for effective health care: immediate steps and strategicdirections. Comput Sci and Telecom Board, Nat Res Council. Washington: National Academies Press.

    Overarching Grand Challenge: Cognitive Support

  • 8/9/2019 Closing Plenary: Bill Stead

    10/24

    Burning Platform: Overwhelming Complexity

    SetsofFac

    tsperDecision

    1000

    10

    100

    5Human

    CognitiveCapacity

    2000 20101990 2020

    Structural Genetics:

    e.g. SNPs, haplotypes

    Functional Genetics:Gene expression

    profiles

    Proteomics and othereffector molecules

    Decisions by ClinicalPhenotype

    Stead WW. Beyond expert-based practice. IOM (Institute of Medicine). Evidence-based medicine and thechanging nature of health care: 2007 IOM annual meeting summary,(Introduction and Overview, p. 19).Washington, DC: The National Academies Press 2008.

  • 8/9/2019 Closing Plenary: Bill Stead

    11/24

    The Idea Take 2

    Work at multiple scales to manage complexity Triangulate multiple signals for robustness

    Satellite

    Doppler Radar

    Rain Gauge

  • 8/9/2019 Closing Plenary: Bill Stead

    12/24

    symptom

    mechanism

  • 8/9/2019 Closing Plenary: Bill Stead

    13/24

    OLD NEW

    One integrated set of dataSets of data from multiplesources

    Capture data in standardizedterminology

    Capture raw signal and annotatewith standard terminology.

    Single source of truthCurrent interpretation of multiplerelated signals

    Seamless transfer amongsystems

    Visualization of the collectiveoutput of relevant systems

    Clinician uses the computer toupdate the record during thepatient visit.

    Clinician & patient work togetherwith shared records andinformation.

    The system provides transaction-level data.

    The system provides cognitivesupport.

    Work processes are programmedand adapt through non-systematic work around.

    People, process and technologywork together as a system.

    Shift EHR Computational Paradigm

    Stead WW. Electronic Health Records. In: Rouse WB, Cortese DA, eds. Engineering the system of healthcare delivery.Tennenbaum Institute Series on Enterprise Systems, Vol. 3. Amsterdam: IOS Press; 2009.

  • 8/9/2019 Closing Plenary: Bill Stead

    14/24

    Data Mining

    Automation

    ConnectivityDecisionSupport

    AggregateEHR

    Diseasemanagement

    dashboards

    Work lists

    Evidence-based

    advisors

    Match Computational Approach toComplexity of Data

    Stead WW. Electronic Health Records. In: Rouse WB, Cortese DA, eds. Engineering the system of healthcare delivery.Tennenbaum Institute Series on Enterprise Systems, Vol. 3. Amsterdam: IOS Press; 2009.

  • 8/9/2019 Closing Plenary: Bill Stead

    15/24

    Use different scales to supporta systems approach to care

  • 8/9/2019 Closing Plenary: Bill Stead

    16/24

    Use different scales to supporta systems approach to care

    Evidence-based Medicine

    Consistent Process

    Visualizationof Results vs. Plan

    IterativeImprovement

    Outcomes

  • 8/9/2019 Closing Plenary: Bill Stead

    17/24

    Use different scales to supporta systems approach to care

    Evidence-based Medicine

    Consistent Process

    Visualization

    of Results vs. Plan

    IterativeImprovement

    Outcomes

  • 8/9/2019 Closing Plenary: Bill Stead

    18/24

    Use different scales to supporta systems approach to care

    Evidence-based Medicine

    Consistent Process

    Visualizationof Results vs. Plan

    IterativeImprovement

    Outcomes

  • 8/9/2019 Closing Plenary: Bill Stead

    19/24

    100

    150

    200

    250

    300

    2005 2006 2007 2008 2009

    Source: UHC and Vanderbilt Data

    3. Mortality for VanderbiltVentilator Patients Compare to

    all the other Hospitals Best in the U.S.

    1. Number of VentilatorAcquired Pneumonia (VAP)

    Cases/Year at Vanderbilt

    Fiscal Year

    2009Results c/w

    2008

    VAPsPrevented 108

    DeathsAvoided 16

    $ Saved $4.3M

    HospitalDays

    Avoided 1055

    ICU DaysAvoided 431

    2. Impact on Results

    Systems Approach to Care

    Vanderbilt now # 1O/E Vent MortalityO/E Length of StayO/E Cost

  • 8/9/2019 Closing Plenary: Bill Stead

    20/24

    Abstraction Generalization

    Observation One Instance

    Modelformal

    relationship

    FeatureSet

    package of

    relatedattributes

    Attributestructured

    information

    raw signalData

    The Idea Take 3

  • 8/9/2019 Closing Plenary: Bill Stead

    21/24

    Generic Interface

    Engine & Core ofInteroperable Data

    Terminology Mining & Management

    ContentBuilders

    Terminology

    Editors

    Populate Operational ToolsHED

    StarFormsRxStarCHISL

    Populate EnterpriseData Warehouse

    Metadata

    External StandardsUMLS

    SNOMEDFDB

    VUMC Terminology Map(Natural Language)

    VUMC Health Language Server(Curated)

  • 8/9/2019 Closing Plenary: Bill Stead

    22/24

    Text String Matching: Ace Inhibitors

    Ac?uprilAc?reticAceon

    AltaceAtacandAvalid*

    AvaproBenazeprilBenicar

    Cand?artanCapotenCapozid*

    ? = wild card character* = accept any terminal ending

  • 8/9/2019 Closing Plenary: Bill Stead

    23/24

    Low Level Facts Combine To ProvideInformation

    Nurse charts milliliters of TPN,

    tube feeding or dextrose

    administered per hour

    Nutritionprovides

    calories and fat

    per milliliter

    Product is carbohydrate

    calories and fat per hour

    Which fuels the statistical

    prediction of hypoglycemia

  • 8/9/2019 Closing Plenary: Bill Stead

    24/24

    Action Steps

    Capture data from any source in any form & archiveas raw signal

    Keep data from different scales of biology or levels

    of abstraction separate

    Aggregate and interpret data purpose by purpose

    Use statistical approaches to separate signal from

    noise & correlate multiple weak signals

    Share algorithms & knowledge sources