Avatars alive! The integration of physiology models and computer generated avatars in a multiplayer online simulation

Abstract

In a mass casualty incident, injured and at-risk patients will pass through a continuum of care from many different providers acting as a team in a clinical environment. As presented at MMVR 14 [Kaufman, et al 2006], formative evaluations have shown that simulation practice is nearly as good as, and in some cases better than, live exercises for stimulating learners to integrate their procedural knowledge in new circumstances through experiential practice. However, to date, multiplayer game technologies have given limited physiological fidelity to their characters, thus limiting the realism and complexity of the scenarios that can be practiced by medical professionals. This paper describes the status of a follow-on program to merge medical and gaming technologies so that computer generated, but human-controlled, avatars used in a simulated, mass casualty training environment will exhibit realistic life signs. This advance introduces a new level of medical fidelity to simulated mass casualty scenarios that can represent thousands of injuries. The program is identifying the critical instructional challenges and related system engineering issues associated with the incorporation of multiple state-of-the-art physiological models into the computer generated synthetic representation of patients. The work is a collaboration between Forterra Systems and the SUMMIT group of Stanford University Medical School, and is sponsored by the US Army Medical Command's Telemedicine and Advanced Technologies Research Center (TATRC).

Full text

Avatars Alive!
The Integration of Physiology Models and Computer Generated Avatars in a
Multiplayer Online Simulation
Laura Kusumoto, Forterra Systems Inc, lkusumoto@forterrainc.com
LeRoy Heinrichs, Parvati Dev, and Patricia Youngblood,
SUMMIT (Stanford
University Medical Media and Information Technologies)
In
a mass casualty incident, injured and at-
risk
patients will pass through a continuum of care
from many different providers acting as a team
in a clinical environment. As presented at
MMVR 14 [Kaufman, et al 2006], f
ormative
evaluations have shown that simulation practice
is nearly as good as, and in some cases better
than, live exercises for stimulating learners to
integrate their procedural knowledge in new
circumstances through experiential prac
tice.
However, to date, multiplayer game technologies
have given
limited physiological fidelity to their
characters, thus
limiting the realism and
complexity of the scenarios that can be practiced
by medical professionals.
This paper describes the status of a
follow
-
on
program to merge medical and gaming
technologies so that computer generated, but
human
-controlled, avatars used in a simulated,
mass casualty training environment will exhibit
realistic life signs. This advance introduces a
new level of medical fidelity to simulated mass
casualty scenarios that can represent thousands
of injuries. The program is identifying the
critical instructional challenges and related
system engineering issues associated with the
incorporation of multiple state-
of
-
the
-
ar
t
physiological models into the computer
generated synthetic representation of patients.
The work is a collaboration between Forterra
Systems and the SUMMIT group of Stanford
University Medical School, and is sponsored by
the US Army Medical Command's Tele
medicine
and Advanced Technologies Research Center
(TATRC).
This program is developing general application
program interface (API) between Forterra
Systems
massively multiplayer simulation
technology, the Online Interactive Virtual
Environment (OLIVE), a
nd
two specific
physiological models. To provide context for the
design of the API, we are
also
developing
curricula for training medical personnel, in
which the models will drive the medical states of
multiple avatars in a virtual environment that
simulat
es
the
mass casualty scenario
s.
To inform the design of the
API, the project team
surveyed physiological models available from
universities and commercial sources, and
identified two basic types for further
investigation, namely rule-based models and
math
ematical models. This paper enumerates the
implications of each type of model on both the
design of the API and on the learning objectives
that
the
resulting system can support.
The paper details
the
design challenges inherent
in visualizing the output of the medical models
in a multiplayer virtual environment, and
provides examples of how these challenges may
be addressed. Specific
design
issues to be
discussed include:
Determining the
optimal
presentation of avatar
symptoms and medical treatments (
e.g.,
spoken word, text, images, representations on
the
3D
avatar
)
Managing the computational load of multiple
physiological models within a computationally
demanding
,
3D online
virtual environment
Mitigating the interactions between medical
models, human control, and artificial
intelligence (AI) control in determination of
avatar state and behaviors.
The results of formative evaluation of the mass
casualty response curricula among hospital
physicians and nurses, using the physiological
models, will be available for presentation at
MMVR 15. They will provide a preliminary
indication of how effectively these models have
been integrated into the avatars, from the
standpoint of instructors and trainees.
Reference
s:
Kaufman, M.;
Heinrichs, L; Youngblood, P
Training
of Medical First Responders for
CBRNE Events Using Multiplayer Game
Technology
, Present
ation at MMVR 14,
January 24
-
27, 2006.