Genetic algorithm with logistic regression for prediction of progression to Alzheimer's disease

BMC Bioinformatics

Table 4 Variable selection from GA for prediction of conversion from HC to MCI/AD, compared with random selections of features.

# of features	Average AUC (GA selected variable sets)	Average AUC (randomly selected variable sets)
3	0.89	0.69
4	0.88	0.71
5	0.89	0.74
6	0.89	0.75
7	0.89	0.77
8	0.89	0.77
9	0.90	0.77
11	0.90	0.79

The best performance is associated with sets comprising 3 to 11 features and can be classified as good-to excellent performance (AUC≈0.9). Random selection of features resulted in poor to borderline predictions (AUC<0.79).

ISSN: 1471-2105