Information Technology Reference
In-Depth Information
Fig. 9.
The collaborative learning facility [9]
Table 1.
Automatic normalization performance after initial training (95% CI). (1): without
approximate matching, (2): with approximate matching. [9].
% Completely
Normalized (1),
N=5000
% Correctly
Normalized (1),
N=100
% Completely
Normalized (2),
N=5000
% Correctly
Normalized (2),
N=100
A
95.7
[95.1-96.3]
100
[96.3-100]
96.9
[96.4-97.4]
100
[96.3-100]
B
88.6
[87.7-89.5]
95.0
[88.8-97.9]
89.9
[89.1-90.7]
95.0
[88.8-97.9]
C
83.1
[82.1-84.1]
99.0
[94.6-99.8]
92.5
[91.8-93.2]
99.0
[94.6-99.8]
D
66.0
[64.7-67.3]
93.0
[86.2-96.6]
69.6
[68.3-70.9]
93.0
[86.2-96.6]
E
73.6
[72.4-74.8]
98.0
[93.0-99.5]
76.8
[75.6-78.0]
98.0
[93.0-99.5]
4 Discussion
The efficient and flexible computerized management of the complex and diverse
information space of clinical medicine remains fundamentally an open research ques-
tion. The mixed record of clinical information system implementations so far tells us
that much work remains to be done. Much effort has been poured into the fully for-
malized, structured approach of information representation. As discussed, this results
in a serious maintenance problem as processes, information and semantics of the
domain evolve and new systems are introduced and interconnected.
The approach suggested here, open information management, builds around natural
language as an expressive, evolving and native clinical information management tool.
Human organizations have successfully utilized natural language for a very long time
to exchange information. The level of expressivity of natural languages is high and
even persons that have never met can understand each other provided they speak the
same language. Natural languages have also mechanisms to introduce new words and
concepts, i.e. to evolve.
Search WWH ::
Custom Search