Usage of Probabilistic and General Regression Neural Network for Early Detection and Prevention of Oral Cancer

Table 4

Performance of PNN/GRNN model for prediction of oral cancer stage.

Estimation parameter	Performance (in %)
	Stage I	Stage II	Stage IV	Stage N0
Accuracy	99.90	86.93	86.83	100
True positive (TP)	0.00	7.90	43.02	35.90
True negative (TN)	99.90	79.02	43.80	64.10
False positive (FP)	0.00	0.00	13.17	0.00
False negative (FN)	0.10	13.07	0.00	0.00
Sensitivity	0.00	37.67	100.00	100.00
Specificity	100.00	100.00	76.88	100.00
Geometric mean of sensitivity-specificity	0.00	61.38	87.68	100.00
Positive predictive value	0.00	100.00	76.56	100.00
Negative predictive value	98.01	85.81	100.00	100.00
Geometric mean of PPV and NPV	0.00	92.62	87.50	100.00
Precision	0.00	100.00	76.56	100.00
Recall	0.00	37.67	100.00	100.00
-measure	0.00	0.547	0.867	1.00