Skip to main content
. 2004 Jul-Aug;11(4):320–331. doi: 10.1197/jamia.M1533

Table 5.

Overall Performance of Different Classifiers for Abbreviations

Precision for Abbreviations ABBR (95% confidence interval) %
FP WS NBL MSL TDLL ODLL
A 2 90.9 ( ± 1.1) 91.8 ( ± 1.1) 90.7 ( ± 1.2) 90.6 ( ± 1.2)
4 91.8 ( ± 1.1) 93.4 ( ± 1.0) 91.9 ( ± 1.1) 92.1 ( ± 1.1)
6 91.6 ( ± 1.1) 94.1 ( ± 0.9) 91.4 ( ± 1.1) 92.0 ( ± 1.1)
8 90.8 ( ± 1.1) 93.9 ( ± 0.9) 91.3 ( ± 1.1) 91.9 ( ± 1.1)
10 90.4 ( ± 1.2) 94.1 ( ± 0.9) 91.0 ( ± 1.1) 91.8 ( ± 1.1)
B 2 91.4 ( ± 1.1) 92.0 ( ± 1.1) 90.9 ( ± 1.1) 90.7 ( ± 1.2)
4 94.5 ( ± 0.9) 94.8 ( ± 0.9) 93.3 ( ± 1.0) 93.0 ( ± 1.0)
6 95.5 ( ± 0.8) 95.7 ( ± 0.8) 93.8 ( ± 1.0) 94.0 ( ± 0.9)
8 96.0 ( ± 0.8) 96.1 ( ± 0.8) 93.9 ( ± 0.9) 94.2 ( ± 0.9)
10 96.3 ( ± 0.7) 96.4 ( ± 0.7) 94.1 ( ± 0.9) 94.3 ( ± 0.9)
C 2 91.7 ( ± 1.1) 92.1 ( ± 1.1) 91.3 ( ± 1.1) 91.0 ( ± 1.1)
4 94.6 ( ± 0.9) 94.7 ( ± 0.9) 93.5 ( ± 1.0) 93.1 ( ± 1.0)
6 95.9 ( ± 0.8) 95.9 ( ± 0.8) 94.1 ( ± 0.9) 94.1 ( ± 0.9)
8 96.3 ( ± 0.7) 96.3 ( ± 0.7) 94.3 ( ± 0.9) 94.4 ( ± 0.9)
10 96.9 ( ± 0.7) 96.8 ( ± 0.7) 94.6 ( ± 0.9) 94.7 ( ± 0.9)
E 2 90.0 ( ± 1.2) 93.1 ( ± 1.0) 91.7 ( ± 1.1) 92.3 ( ± 1.1)
4 94.6 ( ± 0.9) 95.8 ( ± 0.8) 93.9 ( ± 0.9) 94.6 ( ± 0.9)
6 96.2 ( ± 0.8) 96.8 ( ± 0.7) 94.7 ( ± 0.9) 95.4 ( ± 0.8)
8 97.2 ( ± 0.7) 97.4 ( ± 0.6) 95.1 ( ± 0.9) 95.8 ( ± 0.8)
10 97.6 ( ± 0.6) 97.7 ( ± 0.6) 95.3 ( ± 0.8) 96.2 ( ± 0.8)
D NA 84.0 ( ± 1.5) 91.0 ( ± 1.1) 91.7 ( ± 1.1) 92.0 ( ± 1.1)
F NA 98.6 ( ± 0.5) 98.5 ( ± 0.5) 95.7 ( ± 0.8) 96.7 ( ± 0.7)

The machine learning algorithm has four choices: NBL, MSL, TDLL, and ODLL. The feature representation (FP) has six options: A, B, C, D, E, and F, where A, B, C, and E have five different window sizes 2, 4, 6, 8, and 10. The 95% confidence interval for each precision value (p) in the table is p ± 0.5∼1.5%.