Model | Hippocampal input | Cortical thickness input | Combined hippocampal and cortical thickness input | |||
---|---|---|---|---|---|---|
r | RMSE | r | RMSE | r | RMSE | |
ADNI1 | ||||||
Linear regression with lasso | 0.22 ± 0.11 | 8.72 ± 0.81 | 0.56 ± 0.08 | 7.44 ± 0.72 | 0.56 ± 0.08 | 7.42 ± 0.74 |
Support vector regression | 0.23 ± 0.11 | 8.70 ± 0.85 | 0.52 ± 0.08 | 7.68 ± 0.76 | 0.53 ± 0.08 | 7.62 ± 0.78 |
Random forest regression | 0.15 ± 0.10 | 9.27 ± 0.80 | 0.54 ± 0.08 | 7.55 ± 0.76 | 0.54 ± 0.08 | 7.51 ± 0.77 |
APANN | 0.53 ± 0.09 | 7.56 ± 0.76 | 0.51 ± 0.10 | 7.67 ± 0.76 | 0.60 ± 0.08 | 7.11 ± 0.72 |
ADNI2 | ||||||
Linear regression with lasso | 0.14 ± 0.11 | 9.69 ± 0.70 | 0.61 ± 0.07 | 7.77 ± 0.71 | 0.61 ± 0.07 | 7.78 ± 0.71 |
Support vector regression | 0.21 ± 0.10 | 9.75 ± 0.79 | 0.63 ± 0.07 | 7.65 ± 0.68 | 0.63 ± 0.07 | 7.66 ± 0.70 |
Random forest regression | 0.24 ± 0.09 | 9.77 ± 0.76 | 0.58 ± 0.07 | 7.97 ± 0.65 | 0.58 ± 0.08 | 7.97 ± 0.67 |
APANN | 0.52 ± 0.07 | 8.32 ± 0.79 | 0.63 ± 0.07 | 7.58 ± 0.71 | 0.68 ± 0.06 | 7.17 ± 0.71 |
ADNI1 + 2 | ||||||
Linear regression with lasso | 0.12 ± 0.08 | 9.37 ± 0.50 | 0.58 ± 0.06 | 7.71 ± 0.48 | 0.58 ± 0.06 | 7.71 ± 0.48 |
Support vector regression | 0.18 ± 0.07 | 9.39 ± 0.54 | 0.59 ± 0.05 | 7.65 ± 0.42 | 0.59 ± 0.05 | 7.65 ± 0.42 |
Random forest regression | 0.18 ± 0.09 | 9.63 ± 0.61 | 0.57 ± 0.05 | 7.76 ± 0.46 | 0.57 ± 0.05 | 7.75 ± 0.46 |
APANN | 0.54 ± 0.06 | 7.99 ± 0.59 | 0.57 ± 0.05 | 7.79 ± 0.51 | 0.63 ± 0.05 | 7.32 ± 0.53 |
ADNI = Alzheimer’s Disease Neuroimaging Initiative; APANN = anatomically partitioned artificial neural network; RMSE = root mean squared error; SD = standard deviation.
↵* Findings are presented as mean ± SD.