Table 3.
Linear discriminant analysis for first two principal shape components (LDA2) and first 50 components (LDA50) compared with standard remodeling indices (Standard)
| |
|
LDA2 |
LDA50 |
Standard |
|||||
|---|---|---|---|---|---|---|---|---|---|
| ED | ES | ED | ES | EF | EDVI | ESVI | LVMI | ||
| Smoking |
-log(p) |
2 |
21 |
46 |
46 |
11 |
0 |
5 |
4 |
| |
Cohen’s d |
0.13 |
0.43 |
0.67 |
0.66 |
0.30 |
0.02 |
0.20 |
0.18 |
| Diabetes |
-log(p) |
4 |
9 |
49 |
48 |
0 |
1 |
1 |
14 |
| |
Cohen’s d |
0.18 |
0.30 |
0.77 |
0.75 |
0.02 |
0.08 |
0.07 |
0.40 |
| Hypertension |
-log(p) |
12 |
20 |
113 |
101 |
5 |
3 |
0 |
30 |
| |
Cohen’s d |
0.33 |
0.42 |
1.09 |
1.03 |
0.20 |
0.16 |
0.03 |
0.53 |
| Sex |
-log(p) |
19 |
182 |
>200 |
>200 |
62 |
0 |
28 |
18 |
| |
Cohen’s d |
0.41 |
1.44 |
2.34 |
2.23 |
0.78 |
0.02 |
0.51 |
0.40 |
| White |
-log(p) |
9 |
29 |
93 |
96 |
2 |
8 |
1 |
8 |
| |
Cohen’s d |
0.29 |
0.53 |
1.01 |
1.02 |
0.12 |
0.27 |
0.08 |
0.27 |
| Chinese |
-log(p) |
12 |
33 |
79 |
98 |
16 |
7 |
15 |
19 |
| |
Cohen’s d |
0.43 |
0.72 |
1.16 |
1.30 |
0.49 |
0.30 |
0.48 |
0.53 |
| Black |
-log(p) |
2 |
13 |
81 |
67 |
5 |
0 |
3 |
6 |
| |
Cohen’s d |
0.15 |
0.41 |
1.12 |
1.01 |
0.25 |
0.02 |
0.18 |
0.27 |
| Hispanic |
-log(p) |
28 |
44 |
93 |
79 |
0 |
26 |
9 |
23 |
| Cohen’s d | 0.59 | 0.75 | 1.13 | 1.03 | 0.02 | 0.57 | 0.32 | 0.53 | |
For ethnicity each test compares one group with the rest (e.g. white vs. non-white). Significance is quantified by –log(p) (e.g. for p = 0.001, −log(p) = 3). Effect size is measured by Cohen’s d, which can be interpreted as the mean distance between two groups in standard deviations (e.g. males and females were separated by 2.34 standard deviations in the LDA50 ED analysis). Bold-faced numbers highlight the highest separation achieved by PCA vs. standard remodeling indices.