Skip to main content
. Author manuscript; available in PMC: 2021 Jan 15.
Published in final edited form as: Neuroimage. 2020 Oct 16;225:117451. doi: 10.1016/j.neuroimage.2020.117451

Table 3. Training set mean and median absolute error for the models.

The mean and median absolute error for all modeling scenarios for the two cutoffs in the training set.

15 mm3 Cutoff
 50 mm3 Cutoff
Modeling scenario Mean absolute error (in months) Median absolute error (in months) Mean absolute error (in months) Median absolute error (in months)
All Sequences (TS) 2.93 2.10 3.81 2.86
All Sequences (R) 3.35 2.70 4.64 3.62
No QSM (TS) 3.06 2.27 3.78 2.66
No QSM (R) 3.45 2.66 4.50 3.45
No T1w+c 4.24 3.12 5.13 3.78
Conventional 4.27 3.12 4.83 3.62
Demographics 6.12 4.41 6.44 5.00
Mean features (TS) 4.67 3.48 4.83 3.29
Mean features (R) 5.36 4.67 5.98 4.64