Skip to main content

Table 5 Logistic regression analysis to select appropriate models for abnormal tau accumulation prediction

From: Visuospatial memory impairment as a potential neurocognitive marker to predict tau pathology in Alzheimer’s continuum

Modelsa

Classification accuracy (%)

-2 LL

χ2

df

p value

Significant test for -2LL difference

One candidate model

 Model A: RCFT delayed

78.2

150.171

76.472

1

< 0.001

 

Two-candidate model

 Model B: R.HC + L. HC

79.3

161.510

65.133

2

< 0.001

Model A vs. B: p = 0.0009

 Model C: R.EC + L. EC

75.9

162.299

64.344

2

< 0.001

Model A vs. C: p = 0.0005

 Model D: RCFT delayed + L.EC

85.6

123.479

103.163

2

< 0.001

Model A vs. D: p<0.001

Three-candidate model

 Model E: RCFT delayed + L.HC + L. EC

88.5

120.828

105.814

3

< 0.001

Model D vs. E: p = 0.1069

  1. RCFT delayed Rey complex figure test delayed recall score, R. HC right hippocampus volume, L HC left hippocampus volume, L. EC left entorhinal cortex thickness
  2. aVariables were adjusted for age, sex, and education, and ICV adjustment was added for HC volume and EC thickness