Algorithm 1: Proposed

Input: Dataset, Number of features $f_j=f_1,f_2,\dots ,f_m$, Number of data $D_i=d_1,d_2,\dots ,d_n$ Output: prediction accuracy increases

Input the number of features $f_j$  For every feature $f_i\in$ Di    Measure of the Bregman divergence and between $f_i$ and ${\phi }_B$    if $({\phi }_B>\delta )$ then     Project the features as similar    else     Project the features as dissimilar  End if    Select the similar feature subset    Remove the dissimilar feature subset  End for    For each information in ‘$d_1$’ with $f_i$     Perform the regression analysis ‘$R$’      If ($R<0.5$) then        Patient data is classified as ‘low risk’        Else If ($R=0.5$) then           Patient data is classified as ‘medium risk’          Else If ($R>0.5$) then           Patient data is classified as ‘high risk’    End if   Obtain the classification.  End for End