- Customized Care
- Customization of health care products and diagnoses to individual
patients can improve the quality of care. For example, computer
design and manufacturing can be used to create prostheses that are
tailored to an individual, rather than designed to meet a specific
of an average individual within a class. Advances in computation,
including speech recognition, vision, and robotics, can lead to new
types of prostheses that help a wider range of disabilities. In
addition to improving the lives of people with severe handicaps,
this could enable more home care for others and lead to potential
savings in the cost of care.
We have started to see some impact of improved genetic understanding
early diagnosis and treatment. As data from the Human Genome project
becomes more widely available, it introduces new opportunities for
customization based on genetic factors.
Customization need not be based on a set of well-understood
behavior if real-time simulation is included in a diagnostic
setting. For example, using a kind of closed-loop simulation,
with the data from a current patient providing the input,
viability of certain drugs or other treatment could be
tested through simulation before being administered to the
patient.
- Chronic Care
- Home health care of the chronically ill naturally defines a
distributed system. Information between health care providers and
chronically ill patients may include text, audio, images, and other
data. As noted below in the discussion of distributed systems, high
bandwidth may be need out of the patient's home and between small
offices and labs, in addition to large hospitals.
From home health care and computing, the next step is
mobile computing, which gives patients added flexibility. In this
area, improved remote care means lower costs, if a large number of
patients can be treated as out-patients.
Personal medical data is extremely sensitive, so a distributed system
of this kind would need strong guarantees of privacy, protection, and
authentication. The current
health care system, which is still largely based on paper and
disconnected computer systems, has a kind of de facto protection, which
comes from the difficulty of finding information in the system. If
the goals of improved organization and high availability are met,
consumers will demand better protection of their medical data. New
models of privacy and protection may be needed to capture the
emergency "need-to-know" scenarios while providing reasonable levels
of privacy.
- Education
- The use of computing technology in medical education includes
education of the public about health issues, education of medical
students, and continuing education of health care professionals. Many
of the issues here are user-interface problems, and are similar or
identical to those found in other attempts to use interactive computing
in education. The user-interfaces should be friendly, context-driven,
and adaptable to the users. Since existing databases contain some of
the educational information, albeit at different levels of expertise,
the systems should be linked together to provide a uniform framework
for accessing medical information.
One problem in medical education is the training of highly skilled
professionals who perform life-threatening procedures. While not
unique to medicine, this component is missing from most traditional
educational settings. The use of simulation combined with novel
user interfaces may eventually provide a solution to this problem.
Rather than practicing delicate brain surgery or anesthesiology on
patients, the health care professionals could gain experience on
a simulated environment. This leads to the last medical applications
we discussed, which is simulation and modeling.
- Simulation and Modeling
- High performance computing platforms are currently being used for
simulation of individual organs and physiological systems.
Interoperability is the critical next step to modeling interactions
between the physical systems.
For simulations to be useful in clinical settings and medical
education, real-time or near real-time performance is needed. For
diagnosis and treatment, closed-loop simulations should use
measured patient data for initializing and controlling the simulation,
and compute outcome predictions based on a simulated model.
As with most of the other applications discussed here, the hardware
and software need to be reliable as well as fast.