Algorithm 1 Algorithm for model optimization, validation and test performance evaluation
procedure optimizedModel$(Data,actualScore)$, performance$(Data,actualScore)$	▹
for $i\leftarrow 1$ to N do	▹ Perform N times test procedure
$trainingSet\leftarrow$data from all subjects except for ith	▹
$testSet\leftarrow$data from ith subject	▹
for $kernelfunction\leftarrow [linear,quadratic,cubic,gaussian]$ do	▹ tune kernel function
for $kernelScale\leftarrow [0.001-1000]$ do	▹ tune kernel scale
for $boxConstraint\leftarrow [0.001-1000]$ do	▹ tune cost parameter
for $j\leftarrow 1$ to $N-1$ do	▹ Perform $N-1$ times validation procedure
$trainingSet\leftarrow$data from $trainingSet$ except for jth subject	▹
$\left[validationSet\right]\leftarrow$ data from jth subject	▹
$Model\leftarrow train\left(model\right(trainingSet\left)\right)$	▹ train model
$valPrediction\left(j\right)\leftarrow predict\left(model\right(validationSet\left)\right)$	▹ predict
end for
$RMSE\leftarrow \sqrt{\frac{1}{N-1}{\sum }_{j=1}^{N-1}{(valPrediction-actualScore)}^2}$	▹
end for
end for
end for
$[kernelFunction,kernelScale,boxConstraint]\leftarrow min\left(RMSE\right)$	▹
$optimizedModel\leftarrow$ model($kernelFunction,kernelScale,boxConstraint$)
$testPrediction\left(i\right)\leftarrow$ predict($optimizedModel(testSet$))	▹ prediction on test set
end for
$performance\leftarrow$ [r,RMSE,MAE]($testprediction,actualScore$)]	▹ test performance
return $optimizedModel,performance$	▹
end procedure