Skip to main content

Table 3 Model fitting criteria for prediabetic subjects by latent Markov analysis

From: Patterns of changes in serum lipid profiles in prediabetic subjects: results from a 16-year prospective cohort study among first-degree relatives of type 2 diabetic patients

Fitted models

Log Likelihood

BIC

AIC

Number of parametera

Classification. Error

Entropy R-squared

2-State 1-Class

−54,789.4762

109,714.1020

109,616.9523

19

0.0920

0.6052

2-State 2-Class

−54,782.2414

109,728.0850

109,610.4828

23

0.0944

0.5924

2-State 3-Class

−54,782.1874

109,756.4296

109,618.3748

27

0.0944

0.5925

3-State 1-Class

−53,737.2923

107,702.2051

107,538.5846

32

0.1421

0.6681

3-State 2-Class

−53,722.8693

107,737.3773

107,527.7385

41

0.1412

0.6695

3-State 3-Class

−53,717.2883

107,790.2337

107,534.5766

50

0.1427

0.6656

  1. aParameters are those unknown quantities that are estimated during the fitting of LMM