Javabased Dynamic Linked Micromap Plots

1
Java-based Dynamic Linked Micromap Plots
Jim X. Chen, (George Mason University),
jchen_at_cs.gmu.edu,Xusheng Wang, (George Mason
University), xwang1_at_gmu.edu,Daniel B. Carr,
(George Mason University), dcarr_at_galaxy.gmu.edu,B
. Sue Bell (National Cancer Institute),
bellsu_at_mail.nih.gov, andLinda W. Pickle,
(National Cancer Institute), picklel_at_mail.nih.gov
2
What is a Linked Micromap Plot?

• A template for displaying spatially-indexed
  statistical summaries Carr et. al. 1998
• Four key features
• At least three columns of study unit descriptors
• micromaps, labels, and statistical
  graphics
• One of the columns is sorted
• The units are grouped into rows of panels to
  focus attention
• Distinct colors distinguish the units in a group.
  The color for a unit links across its
  descriptors.

3
Project Highlights

• Before
• displayed on a stand-alone system
• static presentation (not interactive)
• Now
• web-based interactive system
• data selection geographic resolution and
  variables
• sorting for any column
• mouse tips, popups, drill down, and other viewing
  options
• more interaction, involvement and understanding

4
Data Resource

• Sample data as the test-bed
• The data is restricted to cervix, lung and
  bronchus, but can be readily extended

5
A Snapshot
http//graphics.gmu.edu/xwang/cancer4/
6
Structure

• Four parallel sequences of panels (columns)
• US or State micromaps
• State or County names
• Summary1 Mortality Rate and Confidence Intervals
  
• Summary2 variable

7
Dynamic Features

• A study unit displays the data of a State or a
  County
• View 41 study units in one frame at a time.
  Display becomes scrollable for more study units
• Show statistical value as a dot and its CI
  (confidence interval) with line segments on both
  sides

8
Sorting

• Each column has a sorting button (except
  micromap)
• An up-triangle icon on button ascending sorting
• An down-triangle icon on button descending
  sorting
• Sorting on a summary the dot curve shows the
  relative relationships among the study units

9
Linking

• The descriptors for a study unit can be linked in
  different ways
• Color
• Blinking for an active study unit (mouse pointer)

10
Grouping

• The study units are grouped to help focus
  attention on a few units at a time
• One group includes five study units.
• Five different colors differentiate the study
  units

11
Coloring Scheme

• Three coloring schemes from a pull-down menu
• Spectral for most users
• Sequential for color blind and black-and-white
  printing
• Divergent for color blind, but not for
  black-and-white printing

12
Micromaps

• Locate study units geographically and reveal
  spatial patterns
• Study units are colored in a group with a chosen
  coloring scheme
• Background and foreground colors categorize study
  units not in the current group
• Background study units a gray color with a white
  outline
• Foreground study units a white or yellow color
  with a black outline

13
Foreground Study Units

• A pull-down menu allows a user to define the
  foreground (as well as the background) study
  units
• Has appeared the foreground study units have
  appeared in the previous micromaps
• Will appear the foreground study units will
  appear in the following micromaps
• Above/below median the foreground study units
  are above the median of the sorted units

14
Micromap Magnification

• Allows a user to zoom in on a micromap for
  studying locations and relationships in detail
• Interactive a study unit with coloring schemes
  (mouse pointer)
• Only the current study unit micromap is magnified
• Left-click the button next to the micromap title
• Right-click a study unit name
• Right-click a study unit dot

15
Automatic Scrolling

• From a micromap or a magnified micromap, if a
  study unit does not appear in the display
• click the right mouse button and the
  corresponding study unit will be automatically
  scrolled into the display

16
Drill-down and Navigation

• Clicking an active link from a high-level LM
  plot, a user can drill down to view the
  corresponding low-level LM plot
• Drill-down also works in the magnified micromap
• A navigation pull-down menu is also provided for
  a user to choose and view any LM plot in any level

17
Overall Look

• When the display cant hold all the study units
• Click the button below the magnification button
• an overview window can show the whole statistical
  curve

18
Different Data Sets

• A Cancer pull-down menu is provided for users to
  download and view different statistical data sets
• When a new data set is downloaded from the web
  server, new display layout may be created to view
  the data

19
Others

• There are many more details

20
A Technical Problem

• Web-based LM plots need to bring the current
  state (chosen values, settings, etc.) of the
  current plot (page) to the next plot
• The Internet provides a stateless programming
  environment that can not transfer states from one
  page to another

21
Solutions

• CGI save the states in the request to the
  web-server
• ASP/JSP directly save the states at the
  web-server side

However, both of the solutions require to write
and run code at the web-server side, which is
complicated and inefficient
22
Our Solution

• One Java applet
• Code embedded in one web page (one html file
  instead of multiple)
• Display all LM plots (different web pages) in one
  applet panel
• Allow saving the states at the client side
• Allow implementing the interaction

23
Data Retrieval

• All map data and statistical data are saved as
  files on the web-server
• Applet code directly retrieves the data from the
  web-server and displays the data in a applet
  panel (web-browser)

24
Conclusions

• This system is a prototype for interactive
  LM-plot information visualization
• This project is supported by NCI/NIH