Skip to main content

Table 3 Model performance in the pooled dataset and after internal-external cross-validation

From: Development of a clinical prediction model for the onset of functional decline in people aged 65–75 years: pooled analysis of four European cohort studies

 

Development of model

Internal-external cross-validation

Apparent performance (in 4 pooled cohorts)

Optimism-corrected performancea (in 4 pooled cohorts)

Development in ELSA, InCHIANTI, LASA

External validation in ActiFE-ULM

Development in ActiFE-ULM, InCHIANTI, LASA

External validation in ELSA

Development in ActiFE-ULM, ELSA, LASA

External validation in InCHIANTI

Development in ActiFE-ULM, ELSA, InCHIANTI

External validation in LASA

n = 2560

n = 2560

n = 2110

n = 450

n = 1424

n = 1136

n = 2090

n = 470

n = 2056

n = 504

Discrimination

 C statisticb

0.719 (0.717–0.721)

0.719 (0.716–0.720)

0.721 (0.719–0.723)

0.698 (0.695–0.700)

0.738 (0.736–0.740)

0.691 (0.687–0.694)

0.713 (0.712–0.716)

0.720 (0.718–0.723)

0.711 (0.709–0.713)

0.740 (0.738–0.742)

Calibration

 Interceptc

0.000 (0.000–0.000)

0.059 (0.047–0.073)

0.000 (0.000–0.000)

0.135 (0.113--0.159)

0.000 (0.000–0.000)

− 0.271 (− 0.293- -0.263)

0.000 (0.000–0.000)

− 0.116 (− 0.144- -0.103)

0.000 (0.000–0.000)

− 0.096 (− 0.112- -0.976)

 Slopec

1.000 (1.000–1.000)

1.053 (1.042–1.065)

1.000 (1.000–1.000)

0.949 (0.931–0.958)

1.000 (1.000–1.000)

0.757 (0.739–0.769)

1.000 (1.000–1.000)

0.966 (0.949–0.981)

1.000 (1.000–1.000)

1.215 (1.191–1.228)

  1. Values are median (IQR)
  2. aOptimism 0.049–0.054, determined by internal validation in bootstrap samples (250 samples with replacement)
  3. bC statistic of 0.50 represents no discrimination and 1.00 represents perfect discrimination
  4. cIntercept of 0 and slope of 1 represent perfect calibration