Table 4 Bland–Altman analysis evaluating agreement between brain age predictions from five algorithms (brainageR, DeepBrainNet, XGBoost, ENIGMA, pyment) for no motion versus low motion scans, and no motion versus high motion scans
From: Examining the reliability of brain age algorithms under varying degrees of participant motion
Raw brain age bland altman analyses | |||
|---|---|---|---|
Algorithms | No versus low motion scans | No versus high motion scans | |
brainageR | Bias | − 3.682 | − 5.410 |
LoA | − 14.182–1.142 | − 24.254–1.418 | |
% within MAD | 48.55% | 34.06% | |
DeepBrainNet | Bias | − 0.721 | − 0.816 |
LoA | − 6.9002–4.7241 | − 6.4133–7.7052 | |
% within MAD | 84.780% | 77.540% | |
XGBoost | Bias | − 1.370 | − 3.577 |
LoA | − 12.389–5.976 | − 17.542–7.398 | |
% within MAD | 65.220% | 44.930% | |
ENIGMA | Bias | − 0.689 | − 1.553 |
LoA | − 10.453–10.932 | − 17.317–12.256 | |
% within MAD | 64.490% | 52.170% | |
pyment | Bias | 0.805 | 0.631 |
LoA | − 2.3343–6.9135 | − 5.8886–7.7195 | |
% within MAD | 87.680% | 85.510% | |