Advances in Artificial Neural Systems

Research Article

Comparison of Artificial Neural Network with Logistic Regression as Classification Models for Variable Selection for Prediction of Breast Cancer Patient Outcomes

Table 4

Variables selections for Logistic Regression and Neural Network approaches.
(a) Disease-free survival.
VariablesNNLR
Invaded nodes100%X
Clinical size stage100%X
Number nodes invaded100%X
SBR grade98%X
Histology98%
Necrosis98%
Oestrogen receptors74%X
Skin infiltrating tumour21%X
(b) Mortality from cancer causes.
VariablesNNLR
Clinical number nodules100%
Progesterone receptor100%X
Number nodes invaded100%X
Clinical size stage98%X
SBR grade98%X
Histology98%
Invaded nodes98%
Skin embolus8%X
(c) Local recurrence.
VariablesNNLR
Number nodes invaded98%X
Lymphatic embolus95%X
Ganglion invaded95%
Necrosis95%X
Oestrogen receptors95%X
Histology48%X
Number tumours15%X
Skin invasion13%X
(d) Metastatic recurrence.
VariablesNNLR
Clinical Size stage100%X
Invaded nodes100%
Progesterone receptors100%X
Number of nodes invaded100%X
SBR grade98%X
Histology98%
Oestrogen receptors85%X