Docstoc

Sample Research Related to Health

Document Sample
Sample Research Related to Health Powered By Docstoc
					                Update on
      Health-Related Research in SIE:



                March 19, 2010

                 Prepared by
               Stephen D. Patek
8/9/2011                                1
                                           Outline
• On-going SIE Activities
      – MINDSET: NLM Training Grant
      – Diabetes Technology
            • Artificial Pancreas & Automation Support for Managing Type 1 Diabetes
            • Spin-off Activities
                 – Intensive Glucose Control in the ICU
                 – Diabetes Technology Center
                 – Metabolic Monitoring
      – Dynamic HER/Sensor Interface (DESI)

• Accelerating the Telemedicine Opportunity: Veterans Administration
  Briefing
      – Briefing
      – Discussion Points & Status

• Follow-up Activities (Xin Yao MS Thesis)
      –    A Case for (Near) Real-Time Remote Monitoring: Remote Holter Monitoring
      –    Channel Model
      –    DESI Generic Design
      –    Cloud Testbed

8/9/2011                                                                              2
                  Acknowledgements
• MITRE Corporation
      – WICAT Project
           • Initial Scoping of DESI
           • Prototype Design and Implementation
           • Continued Development and Evaluation of (i) Sensor/DESI
             and (ii) DESI/EHR Channel Allocation Algorithms
• Lockheed Martin Corporation
      – WICAT Project
           • Performance Implications of Security Mechanisms
           • Continued Development of ECG-specific DESI Functions
           • Cloud Testbed

8/9/2011                                                               3
           Ongoing SIE Activities




8/9/2011                            4
            Uva MINDSET: Medical
      Informatics/Systems Engineering
•
                            Program
                  Trainingfunds:
    ~$3M Grant from the NLM
      – 4 SIE faculty advising 3 postdocs, 9
        graduate students and 7 short-
        term trainees
• Sample Research Areas:
      – Remote monitoring of surgery
        room activities
      – Advanced visualization of clinical
        data – individual and populations
      – Improvement of medication
        delivery processes
      – Development of virtual reality and
        physical training system
      – Modeling the neural basis of touch

8/9/2011                                       5
                        Artificial Pancreas
B. P. Kovatchev, M. Breton, S.D. Patek, S. Anderson, et al. (SEAS and Med School)
                                                             Navigator™ cradle –
                             Freestyle Navigator™
                                                            enables real-time data
                            (Abbott Diabetes Care)
                                                              transfer to the PC
             Study
           Participan
                t




    OmniPod Insulin                                           PC Running UVA-
                                     Attending
   Management System                                           Padova - Pavia
                                     Physician
        (Insulet)                                                Algorithm




                               Frequent YSI Provides
                                   Reference BG
8/9/2011                                                                        6
           Upcoming “Multicenter Trials”
     UVA / U. Pavia / U. Padova / UCSB / Sansum / U. Montpellier
• International trials
  beginning Summer
  2010
• UVA Role
    – Trial & Protocol
      Development
      (Med School)
    – System
      Architecture &
      Integration
    – Safety Layer
• Goals:
    – Scientific
      validation of real-
      time control
      strategies
    – Proof-of-concept
      for field testing of
      safety mechanisms

                                                                   7
      Future Artificial Pancreas Platform
                                                   • Data Link Supports:
            CGM “Receiver”                             – Fine tuning of Control
               (RX/TX)                                   Algorithm
                                                       – Device fault mode
                                                         analysis




                                Wireless Uplinks
                                                           • Sensor failures
                Control                                    • Pump failures
               Algorithm                               – Remote Monitoring
                                                           • Extended alerts

                                                   • Key Issues
           Pump “Transmitter”                          – Control Algorithm
                (RX/TX)                                – Mobility
                                                           • multiple wireless
                                                             interfaces
                                                       – Continuous Operation
                      Battery                              • energy efficiency

8/9/2011                                                                         8
                    APP Spin-off 1:
           Tight Glycemic Control in the ICU
• Stress-induced hyperglycemia in the ICU
      – The trauma associated with whatever causes patients to be in the ICU
        disrupts normal metabolism, leading temporarily to symptoms like
        those Type 2 diabetes, even for non-diabetic patients.
      – (Hypothesized) Risks of Hyperglycemia in the ICU:
            • Infection
            • Decelerated health processes
      – Insulin therapy can be used to prevent severe hyperglycemia.

• Project: ARMY Funding through the JDRF
      – Analysis of burn unit data, relating mortality to quality of control (in-
        range average BG and low glucose variability)
      – ICU Glucose-Insulin Simulator
      – UVA TGC Trial

• MINDSET Seminar
8/9/2011                                                                            9
                   APP Spin-off 2:
           UVA Diabetes Technology Center




8/9/2011                                    10
           Dynamic EHR/Sensor Interface (DESI)




                                          DESI




Premise:
   Body sensor resources are limited and should be dynamically adjusted in response to (i) patient
   needs, (ii) resource availability, and (iii) data requirements of back end users.
8/9/2011                                                                                       11
                                DESI Testbed
    Simulated T1DM Patient
    [Dalla Man et al.]
                                            WiFi
                                                                                         T1DM User
                               Aggregator
                                                                                          Tomcat,
                                                                                          JSP/Servlet




                                                                           BASN Data
                                                      WiFi
                  Bluetooth      JAVA
MATLAB, JAVA                                                           EHR Server

                                                                                         Client
                                                               Mirth                   Application
                                                   WiFi                                     A


                                                                                         Client
                               Aggregator                    Database                  Application
                                                             (MySQL)                        B




                                                                           BASN Data
                   Bluetooth   LabVIEW
 TEMPO Sensor Nodes                                                                    TEMPO User
                                        WiFi
                                                                                          LabVIEW

    8/9/2011                                                                                            12
    Adapting Data Processing and Uplink
            Channel Allocation
    Sensor Interface 1                           DESI / EHR           Uplink Interface 1
                                                  Network
    Sensor Interface 2                            Manager             Uplink Interface 2
                               Data Log              &
                                                 Compressio
    Sensor Interface N                            n Engine            Uplink Interface N



     Sensor / DESI Network                 Core and Complimentary Algorithms
           Manager                                     Modules


•   DESI / EHR Network Manager & Compression Engine
      – Maximizes the clinical relevance of data sent to back-end users
      – Receives utility parameters from the Core and Complimentary Algorithms module (influenced
         in part by user feedback through the uplink interfaces)
      – Assesses uplink interfaces (channels and queues)
      – Chooses compression method (modes or object-by-object) and transmissions to appropriate
         uplink interfaces
      –
8/9/2011 Attempts to maximize expected utility of uplink transmissions                          13
            Accelerating the Telemedicine
                    Opportunity:
           Veterans Administration Briefing
                   B. Horowitz and S. Patek
              John Abrams and Mark Cosby (ALIS)
       John Gingrich, Peter Levin, and Adam Darkins (VA)


8/9/2011                                                   14
         VA-Briefing 1: Why We Are Here
• Research is currently underway at UVA and other places
  to create the components for advanced telemedicine
  systems that can revolutionize remote delivery of health
  services through an infrastructure that can support real
  time interactions between patients and healthcare
  providers
   – focusing especially on wireless sensing technology, signal
     processing, and wireless communication in support of near-
     real time intervention for patients with chronic diseases and
     acute conditions
• We’d like to discuss advancing the transition of
   telemedicine research into practice by developing an
   operational prototype of a full-scale system that would
   allow component researchers to embed early versions of
   advanced components into the experimental
   environment that serves the healthcare community 15
 8/9/2011
VA-Briefing 2: Why We Chose to Meet with
       the Veterans Administration?
• Our belief that development of a new system concept is better done
  through organizations with some hierarchical structure (e.g., VA, DOD,
  DHS) than through the loosely coupled health care sector

• Existing VHA and the Care Coordination / Home Telehealth (CCHT)
  Program provides a basis to build upon.


• We can expand the CCHT goals to include:
      – Real-time remote monitoring, signal processing
      – Ability to detect and mitigate acute care events

• We can select an operational test environment to learn from and
  identify the specific system design needs and corresponding benefits
  that would be driven by real-time interaction


8/9/2011                                                                   16
    VA-Briefing 3: Why Focus on a Real-Time
       Interactive System Infrastructure
•   Important benefits for telemedicine can result from real time interactions
     – Remote real-time management of data collection (content and data rate) based on
       the situation
     – Remote support to patients involved in situations requiring response

•   A broadly used telemedicine system that supports real-time interactions will,
    by necessity, need to have:
     – an open design structure,
     – be dependent on standards, and
     – will be difficult to fundamentally modify as legacy capabilities start to dominate
       decisions on design migration (witness the Internet)

•   If we deploy a system for “health” that does not deal with the more stringent
    challenges of real time intervention, we may dramatically delay or even miss the
    opportunity to realize / build / integrate the necessary infrastructure to support
    real time .
           VA-Briefing 4: Acceleration Is Possible by
              Starting at the System-Level Now




8/9/2011                                                18
   VA-Briefing 5: “Learn As You Go” Systems
          Engineering Methodology*
 • Phase 1 – Test Bed Objectives, Metrics, Prototype
   Definition, Team Formation, Initial Plan for
   Implementation (3 months)
 • Phase 2 – Develop initial test bed at selected
   operational site (9 months)
 • Phase 3 – Collect data for evaluation, including
   decisions and plans for refining and adding to the
   initial capability, including plans for adding new team
   members for next steps (6 months)
 • Phase 4 – Develop next iteration of test bed
 • Etc.
*Horowitz, B.M. and J.H. Lambert, Learn As You Go Systems Engineering, IEEE Trans.
    Systems, Man, and Cybernetics, Part A. 2006.
   Nature of the Discussion & Status
• Main Comments/Questions:
      – Bandwidth and security issues are not the main barrier to
        deployment of real-time remote monitoring.
           • There’s lots of low-hanging fruit to be had in logging personal
             health data and uploading data infrequently.
           • There should be an economic argument for new
             infrastructure/technology.
      – Why is (near) real-time remote monitoring needed?
      – What interventions are enabled by giving health providers
        access to data as it is being collected?
• Status:
      – There has been no follow-up yet.
      – We are structuring ongoing research activities in part to
        respond to the concerns raised above.
8/9/2011                                                                       20
           Follow-up Activities




8/9/2011                          21
    Main Follow-up: Xin Yao’s MS Thesis
• Xin Yao, SIE-MS Candidate
      – Xin was a major force in developing of the DESI prototype that we
        demonstrated in the last advisory board meeting.
• MS thesis:
      – Use Cases
           • Characterization of potential applications of real-time remote health
             monitoring.
           • Development of a specific case.
      – Channel Model
           • Characterization of cellular wireless data services, especially in terms of typical
             (not max) data rates and data rate variability as a stochastic process.
      – Generic DESI
           • Development of a generic DESI data processing and uplink channel allocation
             algorithm.
      – Simulation evaluation of the uplink channel allocation algorithm using
        the channel model developed above.

8/9/2011                                                                                      22
Towards a Case for Real-Time Monitoring
Case: Real-Time Supervision of Holter
          Monitor Patients
• Who
   – People who are referred to an outpatient cardiopulminary monitoring
     service.
• Objectives of Mobile Health Telemetry
   – Provide an extra layer of safety for patients sent home with Holter
     monitors – allow rapid response in case the patient’s condition turns
     out to be more serious than thought at first consultation
• System Architecture
   – Wireless sensors: ECG, accelerometer (e.g. TEMPO), audio/video
   – DESI: data aggregator, alarm system, and channel use optimizer
   – “Multiview” monitoring station with attending staff employed by the
     monitoring service,
       • receiving alerts that raise awareness of high(er) risk patients
       • allowing “drill-down” into high fidelity real-time streaming of ECG, accelerometer
         data
       • allowing real-time emergency response
          Private Healthcare Cloud Testbed

                                        Processing
                                          Nodes




                                                         Database


                                               Website
 Body      DESI
Sensors



                   Internet

                                               Insurance
                              Doctor   Nurse                Others
                                               Company
                                                  Staff
                       Main Features
• The testbed is built on our own computer resources using Ubuntu
  Enterprise Cloud technology.
• The testbed is designed to serve a healthcare community. In our
  case, the players are DESI devices and end users, including doctors,
  nurses and other third-party staffs.
• DESI devices send sensor data to and receive commands from the
  cloud testbed through TCP or SSL protocol.
• Databases are constructed on Elastic Block Storage (EBS) and can
  be accessed from virtual machines (VMs) within the cloud.
• A website, e.g., desi.sys.virginia.edu, is created to provide
  information services for end users. Certain charting toolkit, such as
  SVG or Flash, will be used to display sensor data and analysis result
  in Rich Internet Application (RIA) format.
• Role based access control is applied to the testbed to determine
  who can access to the data and what data can be accessed.
• Depends on the needs, additional security mechanisms, such as
  authentication and secure data transmission, can be enabled in the
  testbed.
                                                                      27

				
DOCUMENT INFO
Description: Sample Research Related to Health document sample