The APACHE system
The Acute Physiology
& Chronic Health Evaluation (APACHE) system was created by medical
researchers and doctors seeking a better way to determine which patients
should be retained in a hospital's intensive care unit when beds became
scarce. The system worked by comparing key statistics for a particular
patient with millions of cases with similar characteristics in order to
project likely outcomes. The system capsulized its analysis into a number
for each patient, called the APACHE score. A higher score meant that the
patient stayed in intensive care; a lower score meant the patient left.
It seems startling
that a single, computer-generated number could even be considered useful
in making such a potentially life-or-death decision, and even more so
that it could carry as much weight as a doctor's personal evaluation.
The reasoning behind the APACHE score is that, while there are some things
the human brain is very good at, it tends to focus too much attention
on two types of occurrences: the recent and the unusual. While that sometimes
serves us very well, it is much better, for decisions like this, to take
into account every relevant case. That is something that, in an unaided
state, the human mind is just not capable of doing.
And the system works.
APACHE III "has been shown to be predictive of in-hospital mortality
rate. It has performed equally well in community based and academic settings.
[But] The APACHE, like other scoring systems presents some limitations.
It is dependent on an operator and requires a long period of training
to be used efficiently” (Murabito, Rubulotta, and Gullo 2008, 349). The
point is not that machines are better than the human mind; the point is
that the mind and the machine together produce the wisest decisions.
Murabito, P., F. Rubulotta,
and A. Gullo. 2008. Quality management in the ICU: Understanding the process
and improving the art. In Anesthesia, pain, intensive care, and emergency:
Proceedings of the 22nd postgraduate course in critical care medicine,
Venice-Mestre, Italy, November 9-11, 2007, ed. A. Gullo, 345-404.
Milan, Italy: Springer-Verlag Italia.