Table 1.
Demographic and clinical characteristics of subjects included in the study across training, validation, and test data sets.
| Total | Machine Learning Datasets | |||
|---|---|---|---|---|
| Training | Validation | Test | ||
| Number of Visual Fields | 29,161 | 23,744 (81%) | 2,577 (9%) | 2,840 (9%) |
| Number of Patients | 3,832 | 3,108 (81%) | 363 (9%) | 361 (9%) |
| Normal (n, %) | 667 (17%) | 547 (18%) | 58 (16%) | 62 (17%) |
| Suspect (n, %) | 2,221 (58%) | 1,793 (58%) | 222 (61%) | 206 (57%) |
| Glaucoma (n, %) | 944 (25%) | 768 (25%) | 83 (23%) | 93 (26%) |
| Number of visits | 7.61 (7.35) | 7.64 (7.36) | 7.10 (7.24) | 7.87 (7.29) |
| Follow-up period (years) | 4.95 (5.25) | 4.94 (5.24) | 4.67 (5.36) | 5.33 (5.24) |
| Age (years) | 60.94 (15.66) | 61.04 (15.77) | 60.38 (15.9) | 60.67 (14.47) |
| Baseline MD (dB) | −3.55 (5.71) | −3.57 (5.72) | −3.76 (6.12) | −3.23 (5.17) |
| Baseline PSD (dB) | 3.42 (3.29) | 3.42 (3.27) | 3.49 (3.41) | 3.42 (3.34) |
The training, validation, and test datasets are for one cross-validation fold.
MD = mean deviation; PSD = Pattern standard deviation. Values are presented as mean (standard deviation), unless otherwise noted.