Optimal Feature Selection-Based Dental Caries Prediction Model Using Machine Learning for Decision Support System

. 2023 Feb 13;10(2):245. doi: 10.3390/bioengineering10020245

Algorithm 1: Proposed model for Dental Caries Prediction.

Input: Training dataset

S : =

(

x_{1}

y_{1}

), (

x_{2}

y_{2}

), …, (

x_{n}

y_{n}

);

F_{n}

Number of features to be selected, and

θ

threshold of feature importance.
Output: highest performing classification approach C.

Begin algorithm:
$f o r (f e a t u r e f_{i}$ in $S$ , $i + 1)$ : Compute the weight importance of features.

relevance =

m u t u a l I n f o r m a t i o n

(

f_{i}

, y)
redundancy = initialize to 0

f o r (f e a t u r e f_{j}

S

j + 1)

r e d u n d a n c y

m u t u a l I n f o r m a t i o n

(fi, fj)
end for

m r m r V a l u e s [f_{i}] =

relevance

- r e d u n d a n c y

;

3.
end for
4.
$H = s o r t (m r m r V a l u e s) . t a k e (θ)$ select a subset of the most important features from the total set S.
5.
$f o r (f e a t u r e f_{k}$ in $H$ , $i + 1)$ : rank features by their predictive power and select the most important one.
- Compute the Gini impurity of the target variable on $f_{k}$ as $G I$ = $G i n i I m p u r i t y (f_{k})$ .
- Split the data samples $x$ from $f_{k}$ into two subsets $x_{1 ~ k}$ and $x_{k ~ n}$ then compute the Avg Gini impurity of the two subsets.
  ${\bar{G}}_{a v g_1} = A V G G i n i I m p u r (x_{1 ~ k})$
  
  ${\bar{G}}_{a v g_2} = A V G G i n i I m p u r (x_{k ~ n})$
- Store in the list of feature importance the Gini impurity caused by the splitting of the data on the current feature.
6.
Normalize the feature importances so that they sum to 1.
7.
Select a final subset of features based on feature importances obtained.
8.
$F e a t u r e s$ = $S e l e c t$ ( $List__{f e a t u r e_i m p o r t a n c e}$ , $F_{n}$ )
9.
PERFORM the Caries Prediction

Hyper-parameter
Machine Learning model include:

List__{C l a s s i f i e r s}

: GBDT, RF, SVM, LR, LSTM
for

C_{i}

List__{C l a s s i f i e r s}

C_{i} = R a n d o m i z e d S e a r c h A l g o r i t h m (H y p e r_p a r a m e t e r s

)

C_m e t r i c = E v a l u a t e (F e a t u r e s, C, A c c u r a c y, F 1_{s c o r e}, R e c a l l, P r e c i s i o n, R O C

)

L i s t_{C l a s s i f i e r_e v a l u a t i o n_m e t r i c s} . a p p e n d (C_{i}, C_m e t r i c)

end for

10.
Final classifier is obtained as: $C l f = A r g M a x {L i s t_{C l a s s i f i e r_e v a l u a t i o n_m e t r i c s}}$

argmax operator returns the classifier that maximizes the evaluation metric.
11. END algorithm