Table 2.

Model fit statistics for latent class analysis of inflammatory profiles.

No. of classes	AIC	BIC	Log-likelihood	Entropy	BLRT LR	BLRT P	Class proportions, %
2	7,081.4	7,229.3	−3,503.7	0.9556	-	-	34.3/65.7
3	6,815.1	7,038.9	−3,351.5	0.9288	304.37	< 0.001	29/29.8/41.3
4	6,668.4	6,968.1	−3,259.2	0.9216	184.66	< 0.001	15.5/25.4/29.1/30
5	6,591.5	6,967.2	−3,201.8	0.9246	114.87	< 0.001	11.4/14.6/19.5/27.2/27.2

AIC, Akaike information criterion; BIC, Bayesian information criterion; BLRT, bootstrap likelihood ratio test. Indicators included: WBC, CRP, PCT, NLR, PLR, SII, temperature, duration of symptoms, and age (tertile-categorized). Although the 5-class model demonstrated the lowest AIC/BIC values, the 3-class model was selected as the final solution based on highest entropy (0.929), clinical interpretability, and avoidance of over-fragmentation (see Section 3.2).