Title: Internet Health Monitors: Distributed Infrastructure for Measuring Population Health
1Internet Health MonitorsDistributed
Infrastructure for Measuring Population Health
Bruce R. SchatzDepartment of Medical Information
ScienceCollege of Medicine, Institute for
Genomic Biology University of Illinois at
Urbana-Champaign schatz_at_uiuc.edu ,
www.canis.uiuc.edu
Biomedical Informatics Campus Seminar University
of Illinois at Urbana-Champaign October 7, 2005
2Healthcare Infrastructure
- Infrastructure is the Whole System
- Hospital, Clinic, Home
- Doctors, Nurses, Brochures, Internet
- NO Viable Model for Health System
- Too much Cost! Too Much Volume!
3Health Systems
- Healthcare is THE Economic Issue
- BIGGEST item in modern societies
- Fast growing due to aging population
- Healthcare will BREAK Every Nation
- Nations go bankrupt and People die!
4The Fundamental Cause 1
- Medicine versus Health
- Cure Sick in BIG Hospital
- Maintain Wellness in small Clinic
- Recent Rise of Chronic Illness
- No cure with drugs surgery
- only manage with diet exercise
5The Fundamental Cause 2
- Health Systems now for Acute Illness
- Hospitals are Profitable Business but
- Clinics are supported by Government
- Systems cannot handle Chronic Illness
- Chronic Illness now dominates Costs and Systems
cannot handle Volume
6The Viable Solution 1
- Independent Clinics are Doomed
- 1990s America -- small Clinics failed
- when Government support reduced.
- Health Systems start HMOs
- Health Maintenance Organizations
- 2000s America HMOs all failing.
- 2000s Japan small Clinics will fail
- when Government support reduced.
7The Viable Solution 2
- Need Complete Provider Pyramids
- High Level for High Quality at High Cost
- Low Level for Low Quality at Low Cost
- Handle Volume by Pushing Cases Down
- Bottom Levels handle MOST CASES
- Viable Healthcare Infrastructure
- Hospitals with Doctors for Surgery, Clinics with
Nurse for Drugs, Homes with Patients for nearly
all Health Interactions!
8Health Informatics
- Need New Viable Infrastructure
- Health Information Technology
- Provides Support for Patients in Homes
- Creates Bottom of Pyramid to Offload
- Informatics can Solve this Problem
- Patients themselves create population health
database via informatics that automatically
routes healthcare
9Informatics Technologies
- Measure Population Health
- Adaptive Question Asking of Quality of Life
Questionnaires - Answers for Individuals creates Database for the
Population - Manage Population Health
- Structured Health Vectors from normalized patient
records - Statistical Information Retrieval cluster
patients into care cohorts
10Measure Population Health 1
- Quality of Life Questionnaires
- Self-Assessment directly by Patients
- General Status questions, e.g. SF-36
- Specific Disease questions, e.g.
- Arthritis Can you walk without pain?
- Heart Disease Do your ankles swell?
- QoL correctly does coarse prediction
- VA Heart Study SF-12 better than surgeon about
patient survival
11Measure Population Health 2
- Electronic Records for fine prediction
- Paper supports 10s of questions
- Electronic supports 100s or 1000s
- Adaptive Question Asking
- Choose questions by weighted treewalk
- Each session asks 10s of questions customized to
particular condition - Generate Population Database
- Daily individual records from all homes
12Manage Population Health
- Structured Health Vectors
- Patient answers Questions daily
- Average scores generate Health Vector
- Elements of Vector are Meaningful
- Cluster Patient Cohorts
- Normalize Vectors for Similar Clusters
- Weight Question Groups Medically
- Route Care into Pyramid using Clusters to
Determine Cohorts
13Theory Experiment
- Questionnaire from Merged QoL
- 120 questions from 20 questionnaires
- General plus some Specific questions
- Simple Clusters do coarse prediction
- Students simulate sick or well patients
- K-means with random seeds does correct clustering
from actual health monitor sessions with 100
answers
14Practice Experiment
- Practical Risk Assessment Possible?
- Need 4 Cohort Clusters correctly predicted
hospital, clinic, telephone, home - Is 120 questions (10more) enough?
- What Clusters can do fine prediction?
- Use Historical Database of Real Patients
answering Paper QoL Questionnaires - Agglomerative with complete link always
consistent but always correct? May need
appropriate structured vector weighting
15Clinical Experiment
- Real Patients in Real Settings
- 1000 senior patients with heart disease
- Use in Medicare Coordinated Care
- Telephone Interface via voice response
- Determine Care Levels automatically
- Demonstrate Feasible Technologies
- Adaptive Question Asking with Faceted Category
Classification - Statistical Cohort Clustering with Structured
Vector Weighting
16Current Prototype Status
- Theory Experiment
- Completed in LIS Healthcare Infrastructure
- Health Informatics Information Retrieval
- Practice Experiment
- On-going collaboration with Carle Hospital
- Adaptive Faceted being Developed
- Historical Databases being Analyzed
- Clinical Experiment
- Multi-Year Trial Proposal to US AHRQ
Agency for Healthcare Research Quality
17Sample General Health Questions
18Sample Specific Health Questions
19Health Monitor Session
20Beyond Screening
- Why are Some People Healthy? (R. Evans)
- Major categories are disease, health care,
health function, genetic endowment, physical
environment, social environment, individual
response, behavior, well-being, prosperity. - Healthy People 2010
- 467 objectives in 28 focus areas
- www.health.gov/healthypeople
- Measure Full-Spectrum Health Status
- Detailed QoL in each detailed category
21Computer-based Questionnaires
- Treat actual disease stage (dynamic)
- Computer assessment handles full-spectrum
- Database of all questions (300K)
- Individual session asks only 30 questions
- Tree-walking Categories by Breadth-First
- Treat actual patient status (adaptive)
- MOS knows this the problem (McHorney)
- GRE as the paradigm
- Session answers determine questions
- Historical answers determine questions
22Population Management
- Possible to Monitor Whole Populations
- Daily Monitors, Full Spectrum of Features
- Internet Software handles Questionnaires
- Cohort Clusters supplement Diagnoses
- Daily Feature Record for each Individual
- Detailed Databases for whole Population
- Analyze Clusters of Similar Patients
- Cohort Switching drive Treatments
- Manage Expectations with Actual Cases
- Improve Health by Switching Cohorts
23Getting from Here to There
- Develop Full-spectrum Questionnaire
- Merge existing Quality of Life instruments
- Encode knowledge from Medical Professionals
- Develop Dynamic Adaptive Administration
- Software to handle Interactive Sessions
- Software to build Individual History
- Software to build Population Database
- Develop Cohort Similarity Clustering
- Algorithms for Statistical Feature Matching
- Lifestyle Coaching via Cohort Switching
- Deploy Test (10) to Trial (1000) Population
24Healthcare Infrastructure
- Provider Pyramids
- Scale to Volumes for Chronic Illness
- Risk Assessment
- Automatically Determine Level of Care
25Further Information
- Papers
- See articles on Internet Health Monitors and on
Monitoring Population Health by R. Berlin, MD,
and B. Schatz, PhD, at www.canis.uiuc.edu under
Publications under Papers. - Demos
- Try prototypes and view analysis at
www.canis.uiuc.edu under MedSpace at bottom of
web page.
26Further Reading
- Richard Berlin and Bruce Schatz (2001)
- Population Monitoring of Quality of Life for
Congestive Heart Failure, Congestive Heart
Failure, 7(1)13-21 (Jan/Feb 2001). - Colleen McHorney (1997)
- Generic Health Measurement Past Accomplishments
and a Measurement Paradigm for the 21st Century,
Annuals Internal Medicine,127743-750. - R. Evans, M. Barer, T. Marmor (eds)
- Why are some People Healthy and Others Not?
- The Determinants of Health of Populations
- (New York Aldine de Gruyter, 1990).
- G. Rose, The Strategy of Preventive Medicine
- (Oxford University Press, 1992).