Table 1 Discriminant models data.

(a) Data suitability for discriminant analysis
Pooled within-groups matrices (correlation between predictor variables)						Box’s M test p value (cutoff > 0.001)
	K⁺	Na⁺	LDH	MRC1	GST	Box’s M test p value (cutoff > 0.001)
K⁺	1	−0.065	−0.215	−0.016	0.12	0.015
Na⁺	−0.065	1	0.136	−0.038	−0.03
LDH	−0.215	0.136	1	−0.129	−0.103
MRC1	−0.016	−0.038	−0.129	1	0.049
GST	0.12	−0.03	−0.103	0.049	1

(b) Model fitness
Wilk’s Lambda	p value
0.525/0.545	2.66 × 10^–9/2.24 × 10^–9

(c) Variable contribution to the discriminant model and inter-group difference of means
	SCDFC	Wilks’ Lambda (equal for both models)	F test p value (cutoff < 0.01) (equal for both models)
K⁺	−0.626/0.661	0.732 (26.8%)	8.66 × 10^–7
GST	0.529/−0.547	0.876 (12.4%)	0.0013
MRC1	−0.392/0.414	0.871 (12.9%)	0.0011
LDH	0.294/−0.341	0.839 (16.1%)	0.0002
Na⁺ (excluded stepwise)	0.276/NC	0.898 (10.2%)	0.0038

(d) Discriminant function discriminatory power
	Eigenvalue	Canonical correlation
First discriminant function	0.904/0.837	0.689/0.675

(a) Test results aiming to evaluate data suitability for discriminant analysis. (b) Model significance and fitness indicators. (c) Indicators of importance of individual variables to the model. SCDFC indicates the contributions made by each variable to the model, Wilks’ Lambda indicates the amount of variance that is unrelated to group membership with the percentage of relevant variance in parentheses, and the F test p value indicates the significance of differences between group means for each variable. (d) Canonical discriminant function ability to discriminate between groups. The higher the eigenvalue and canonical correlation, the better the model’s ability to discriminate between groups. Results of the all-inclusive and stepwise models are shown separated by a forward slash in (b), SCDFC column in (c), and (d).
SCDFC standardized canonical discriminant function coefficient, NC not calculated.

Search