Diabetes Detection Models in Mexican Patients by Combining Machine Learning Algorithms and Feature Selection Techniques for Clinical and Paraclinical Attributes: A Comparative Evaluation
Table 1
Summary of the results obtained by the related works.
Authors
Feature selection
Results of the evaluated metrics
Chou, Hsu, and Chou
No
(i) Accuracy: 0.80 to 0.95 (ii) Precision: 0.73 to 0.92 (iii) Recall: 0.62 to 0.93 (iv) score: 0.67 to 0.92 (v) AUC: 0.87 to 0.99
Kangra and Singh
No
(i) Accuracy: 0.66 to 0.98 (ii) Precision: 0.65 to 0.98 (iii) Recall: 0.65 to 0.98 (iv) AUC: 0.64 to 0.99 (v) MCC: 0.30 to 0.97 (vi) Kappa value: 0.29 to 0.98
Alcalá-Rmz, Zanella-Calzada, Galván-Tejada et al.
No
(i) Accuracy: 0.94 to 0.98 (ii) Loss function: 0.19 to 0.25 (iii) AUC: 0.98
Hu, Li, Lu et al.
Multiview subspace clustering guided
(i) AUC: 0.82
Haq, Li, Khan et al.
Decision tree (ID3), AdaBoost, and random forest
(i) Accuracy: 0.98 to 0.99 (ii) Recall: 0.98 to 1.00 (iii) Specificity: 0.97 to 0.99 (iv) Sensitivity: 0.98 to 1.00 (v) Precision: 0.99 to 1.00 (vi) MCC: 0.97 to 0.99 (vii) score: 0.98 to 1.00 (viii) ROC curve: 0.98 to 0.99
Sneha and Gangil
Correlation value
(i) Specificity: 0.98 (ii) Accuracy: 0.82
Olabanjo, Wusu, and Mazzara
Ensemble feature selection
(i) score: 0.69 to 1.00 (ii) Recall: 0.69 to 0.92 (iii) Precision: 0.83 to 1.00