Table 7.
Validation of the grading model proposed by Feldman et al. [27] and this work with DCCT/EDIC ground truth using Fisher exact test.
| DCCT/EDIC ground truth | Total no of samples in the dataset | ||
|---|---|---|---|
| Non-DSPN | DSPN | ||
| Severity grading by Feldman et al. [27] | |||
| Absent | 2610 (100%) | 0 (0%) | 2610 |
| Mild | 219 (21%) | 815 (79%) | 1034 |
| Moderate | 0 (0%) | 1092 (100%) | 1092 |
| Severe | 8 (0.67%) | 1194 (99.33%) | 1202 |
| Total | 2837 | 3101 | 5939 |
|
| |||
| Severity grading by this work | |||
| Absent | 2568 (98.4%) | 42 (1.61%) | 2610 |
| Mild | 52 (5.03%) | 982 (94.97%) | 1034 |
| Moderate | 0 (0%) | 1092 (100%) | 1092 |
| Severe | 0 (0%) | 1202 (100%) | 1202 |
| Total | 2837 | 3101 | 5939 |