CMSC 838B: Zoomable User Interfaces
Nizar Habash
Summary and Analysis
LifeLines: Using Visualization
to Enhance Navigation
and Analysis of Patient Records
Plaisant, C., Mushlin, R., Snyder, A., Li, J.,
Heller, D., & Shneiderman, B.
Summary of Contributions
This paper addresses the issue of providing a medical history
visualization that facilitates, expedites, and improves doctors'
ability to review their patients' records. The LifeLines project
uses the intuitive concept of a timeline where events appear as
dots or lines on a 1-D temporal grid as the basis for the
visualization. This basic concept is coupled with many other
features rendering the LifeLines display practically a large menu
that orders, relates, and provides access to large amounts of
information. Some of these features include:
- Display grid: Related events (diagnoses,
medications, etc.) are displayed together in groups
called facets. Within these facets, similar events such
as migraine headache diagnosis are displayed on
common lines. This grouping, together with the timeline
display, highlights relationships between events and
allows a first-glance access to a wealth of information.
To deal with the problem of data crowding, facets can be
closed to provide more screen space for viewing other
facets without having to scroll.
- Details on demand: Moving the cursor
over an event on the LifeLines interface causes detailed
information about the event to be displayed in a textbox
on the upper right corner of the interface.
Double-clicking on some events causes related documents
to be displayed in a separate window. For example, moving
over an event labeled as x-ray displays the date
and time when the x-ray was taken; while double-clicking
on the event will display the x-ray itself.
- Zooming: Zooming in and out allows
accessing details and viewing the whole context when
needed. Zooming is controlled either by the powerful
zoom-and-pan slider or by clicking on the background
(i.e. left button zooms in, right button zooms out).
- Highlighting relationships: When a
search is conducted, the results are highlighted in all
the related facets.
- Coding attributes: Label, color, and
line thickness are used to emphasize certain events or
attributes.
- Outlining: When facets are closed, a
silhouette showing an outline of the facet's contents is
displayed.
- Summarizing: Events can be automatically
aggregated and replaced by a summary event when the user
zooms out.
A Comment on the Data Structure:
The LifeLines interface was developed with Java and it uses an
Access database. The interface is designed to be general enough
for a number of applications (personal medical history, personal
shopping history, etc.) and a variety of user needs. Data
administrators set the default profiles for the application, but
users have the ability of mapping display attributes such as
color and font size to data attributes.
Related Web Sites
Human
Computer Interface Lab: LifeLines Project Home Page
LifeLines
Project with the Department of Juvenile Justice
LifeLines
Demo
Comments
Contributions
- The paper is well written and had a number of good
examples.
- The interface seemed overall intuitive and easy to use.
The amount of information displayed in the medical
history demo is impressive, allowing to see how events
relate to each other within facets, by color grouping,
and in time. Also, the ability to load documents of
various types in the window next to the LifeLines
interface is very useful.
- This paper provided good examples on interface problem
solving, especially regarding data crowding.
Issues
- The LifeLines demo did not run on a Sun workstation using
Netscape 4.0, but it ran on Internet Explore under
Windows. Also, the speed of response time could be better
(screen refreshing especially). These could be Java
problems.
- In all the demos shown, the amount of data was large, but
for a small period of time (maximally 3 years). I would
like to see how the interface can handle large amounts of
data over long periods of time.
- One annoying aspect of the interface is the overloading
of functionality of the mouse buttons based on the
cursor's position on the interface. This would have been
tolerable if the cursor changed shape to prompt the user
to the buttons' available functionality. For example, the
cursor looks like a pointing hand only over events with
linked documents; it appears as a magnifying glass when
it is over the background (where zooming is allowed); and
then, in all other areas, as a plain pointer.
Questions to the Authors
- In the paper, the authors mention that they have begun
testing the LifeLines medical history project on real
data. What kind of experiments have they run? What are
the results?
- Do the authors envision a way where information from
different LifeLines applications can be used together?
For example, overlapping medical history and shopping
history can provide information on reasons behind
allergies or other problems. Does the architecture of the
LifeLines interface allow for this idea to be implemented
in its general case?
Questions
- Have you seen the LifeLines demo? What do you think of
it? Do you agree/disagree with the issues brought up
above?
- The LifeLines project has been used to display criminal,
medical, and shopping histories. Can you think of other
applications where LifeLines can be utilized?
- In the medical history demo, each event usually had one
or no document linked to it. How do you suggest providing
access to multiple documents?