Table 2 Model performance when trained on different views.
From: Confounders mediate AI prediction of demographics in medical imaging
Sex (AUC) | Age (MAE) | White vs Rest (AUC) | Black vs Rest (AUC) | Asian vs Rest (AUC) | |
|---|---|---|---|---|---|
Cedars-Sinai | |||||
Apical 4 chamber | 0.84 (0.84–0.85) | 9.66 (9.55–9.77) | 0.59 (0.58–0.60) | 0.60 (0.59–0.61) | 0.54 (0.52–0.55) |
Apical 2 chamber | 0.80 (0.79–0.81) | 10.82 (10.61–11.03) | 0.58 (0.57–0.60) | 0.60 (0.58–0.61) | 0.55 (0.53–0.58) |
Parasternal long axis | 0.84 (0.83–0.85) | 9.11 (8.97–9.25) | 0.63 (0.61–0.64) | 0.62 (0.61–0.64) | 0.58 (0.56–0.60) |
Subcostal | 0.74 (0.73–0.75) | 11.73 (11.62–11.81) | 0.55 (0.54–0.57) | 0.55 (0.53–0.57) | 0.55 (0.53–0.58) |
Ensemble of all views | 0.93 (0.92–0.94) | 7.78 (7.55–8.0) | 0.71 (0.69–0.74) | 0.72 (0.69–0.74) | 0.60 (0.56–0.64) |
Stanford | |||||
Apical 4 chamber | 0.93 (0.92–0.93) | 7.40 (7.28–7.53) | 0.71 (0.70–0.73) | 0.74 (0.71–0.76) | 0.73 (0.71–0.74) |
