Table 2.
Difference between CLSA subset and similar works aiming to predict cognition related measurements.
| Characteristics | Dataset | |||
|---|---|---|---|---|
| CLSA (ours) | Venugopalan31 | Thanh Duc64 | Oyama65 | |
| Modality | Multimodal (Fundus and metadata) | Multimodal (MRI, EHR, SNP) | rs-fMRI | NIRS |
| Number of images | 25,737 | 202 | 331 | 202 |
| Age (years) | 60.9 (9.55) | N/A | 74.3 (4.70) | 73.4 (13.0) |
| Sex (% male) | 50.66% | N/A | 49.54% | 43.07% |
| Predicted variable(s) | Global cognition, inhibition, memory, speed, executive function, sex, APOE4, BMI, SBP, DBP and age | AD status | MMSE scores and AD status | MMSE scores |
Values from cohorts other than CLSA are reported from corresponding papers. Mean and standard deviation are reported for numerical values.
Significant values are in bold.