Fuzzy Intelligent System for Patients with Preeclampsia in Wearable Devices

Table 5

Results of the methods proposed for classifying preeclampsia patients.

C4.5 features	Accuracy	Relative size	Advantages	Weak points
(i) No pruning; (ii) No limit in number of nodes	0.9945	100%	(i) Good accuracy	(i) Low interpretability because of the tree size; (ii) Overfitting; (iii) Difficult extrapolation out of the dataset
(i) Postpruning (confidence factor = 0.25); (ii) Online pruning (number of instances/node ≥ 2)	0.8216	39%	(i) Encouraging accuracy; (ii) Increased interpretability due to the reduction in the tree size	(i) Difficult extrapolation out of the dataset; (ii) Dependence on measuring instruments; (iii) Not good accuracy
(i) Fuzzy linguistic representation; (ii) Postpruning (confidence factor = 0.25); (iii) Online pruning (number of instances/node ≥ 2)	0.7503	32%	(i) Increased interpretability due to the linguistic approach; (ii) Flexibility on measuring instruments; (iii) Human observation of attributes	(i) Loss of precision