Palm-based Wireless Computing to Improve Access to Health Information and Coordination of Care

1
Palm-based Wireless Computing to Improve Access
to Health Information and Coordination of Care

• James J. Cimino, M.D
• Department of Biomedical Informatics
• Columbia University
• New York Presbyterian Hospital
• Broward General Medical Center Grand Rounds
• July 7, 2005

2
What are we trying to fix?

• Errors!
• Information-related causes of errors
• Lack of access to patient information
• Lack of access to health knowledge
• Poor communication and coordination of care

3
How do we study causes of errors?

• Surveys
• Focus groups
• Observation
• Use of clinical information systems
• Observation with think aloud
• Logfile analysis

4
What did we find?

• Information needs are 1/3 each
• Patient information
• Institutional or administrative information
• Health knowledge
• Information needs and resources are
  context-specific
• Information needs are frequently unresolved
• Team members cant identify each other
• Communication is delayed and inefficient

5
Proposed Solution

• A wireless, handheld device that
• Connects to the clinical information system
• Provides local knowledge resources
• Provide context specific on-line knowledge
• Supports communication among team members
• Does not depend on infrared or wireless Ethernet

6
Where We Are
7
Where We Want to Go
8
Hand-Held Devices Choices

• PalmOSPhone
• Kyocera, Samsung, Handspring

9
Palm Platforms
10
Client Application Choices

• Platform dependent
• PalmOS SDK
• Platform independent
• Web-based
• WCA, AvantGo, EudoraWeb, Internet Explorer

11
Wireless Communication Choices

• Wireless modem
• External modem
• Integrated modem
• Expansion card slot
• Wireless network card
• Wireless LAN
• IEEE 802.11
• Wireless WAN
• Omnisky (EarthLink), Palm.Net
• SprintPCS, Verizon

12
Wireless Handheld Solutions Advantages

• Mobile
• Personal
• Replace multiple devices
• Replaces paper resources
• Nomadic data capture

13
Wireless Handheld Solutions Disadvantages

• Screen size
• Input methods
• Connection time
• Transmission speed/quality
• Local storage
• Security
• Device-specific thick clients
• Limited Web-browsing
• Device cost
• Wireless service cost

14
New York Presbyterian HospitalClinical
Information Systems Architecture
15
WebCIS

• Web-based front end to legacy database
• Medical Entities Dictionary (MED)
• Translation
• Class-based queries
• Patient ? Department?Report header?Details
• Summary reports
• Links to health resources
• Infobuttons
• Patient lists

16
(No Transcript)
17
New York Presbyterian HospitalClinical
Information Systems Architecture
18
PalmCIS Architecture


19
PalmCIS Client

• Kyocera QCP 6035 Smartphone or Treo 650
• EudoraWeb browser
• SprintPCS wireless service
• End-to-end SSL encryption

20
PalmCIS Server

• Apache server
• Thawte SSL Web server certificate

21
PalmCIS Server Application

• Web-based
• C, CGI, HTML
• User interface pull-down menus, links
• Security authorization, authentication, signon
  management, session management, logging
• Data Access Modules (DAMs) for patient data
• Medical Entities Dictionary (MED) for translation
• Patient-list?Department?Report headers?Details

22
Security

• End-to-end Secure Socket Layer
• ID-specific IP address
• Host IP address restrictions
• ID and Password
• SecurID
• Timeout
• System logs
• Login failures generate e-mails
• Screen lock (potential)

23
PalmCIS Login
24
PalmCIS Patient Selection
25
PalmCIS Laboratory Results
26
PalmCIS Radiology Results
27
PalmCIS Summary Reports
28
PalmCIS Microbiology Reports
29
PalmCIS Cardiology Reports
30
PalmCIS etc
31
Information Resources
32
Infobuttons

• Information needs arise while reviewing clinical
  information
• Needs can be predicted based on context
• Task
• User
• Patient
• Patient data
• Queries to on-line information resources can
  resolve needs

33

• Match concept of interest to class(es)
• Identify context based on class(es)
• Retrieve question(s) for context
• Sort by priority
• Return HTML page of links (URLs)

34
PalmCIS Infobutton Culture Result
35
PalmCIS Infobutton Sensitivity
36
Navigational Issues

• Trade-off data per page versus click-to-data
  ratio
• Minimum required information versus fewest clicks
• Details?Department?Header?Details is inefficient
• Can previous path predict future path?
• Log files used to identify most-common paths
• PalmCIS keeps track of users path
• Provide list of shortcuts to next likely report

37
PalmCIS Shortcuts
38
Impact of Shortcuts

• Improve efficiency

• Reduce back clicks

• Optimize traversal paths

39
Impact of Shortcuts

• Improve efficiency

• Reduce back clicks

• Optimize traversal paths

• Benefits
• Facilitate access to information
• Assist in answering patient-specific questions
• Serve as reminders of what to look at and patient
  problems

40
Virtual Whiteboard

• Identify team members and roles
• Permit messaging between team members
• Asynchronous communication
• Palm-based and desk-based

41
(No Transcript)
42
(No Transcript)
43
Cost

• Treo 650 500
• Sprint PCS 60/month
• Annual cost 970/user
• Server equipment extra
• Personnel extra

44
Research Issues

• Usability study
• PalmCIS
• Shortcuts
• Infobuttons
• Understanding whiteboard functions
• Structured note entry
• Where are handhelds better than desktops?
• How do we study usefulness and impact?
• Usage logs
• Increased access to information resources
• Faster access to clinical data and alerts

45
Next Steps

• Redesign of clinical data presentation
• Create efficient, task-specific views
• Based on experience to date
• Complete integration of Virtual Whiteboard and
  note writing with WebCIS
• Explore use of Wi-Fi
• Design randomized controlled trial
• Get funding for larger deployment

46
Conclusions

• Secure, usable, wireless, handheld access to
  patient information and health knowledge is
  feasible
• Cost is not a big factor if replacing cell phone,
  PDA and beeper
• Integration into workflow and educational
  processes is biggest challenge
• Understanding appropriate niche functions is
  critical to success
• Answers are not obvious - research is needed

47
Acknowledgments

• Elizabeth Chen, PhD
• Jianbo Lei, MD, MA
• Larry McKnight, MD
• National Library of Medicine Training Grant