You are viewing the site in preview mode

Skip to main content

Table 1 Parameter selections for further simulation and investigation of the impact of data quality metrics on risk prediction (In bold: Reference values)

From: Joint models in big data: simulation-based guidelines for required data quality in longitudinal electronic health records

Parameter

Parameter Annotation

Parameter Choices

Sample Size

N

\(\{50, 200,\textbf{ 500}, 5000\}\)

Noise Standard Deviation

\(\sigma _\epsilon\)

\(\{0.05,0.075, \mathbf { 0.15}, 0.3\}\)

Percentage of Patients Responding

\(p_{perc}\)

\(\{0, 0.2, 0.5, 0.8, \textbf{1}\}\)

Years of Assumed Slope

\(t_m\)

\(\{1,\textbf{3},5\}\)

Number of Measurements per Year

\(n_{abs}\)

\(\{1,\textbf{2},3\}\)

Intercept Difference

\(\Delta _b\)

\(0, \{\mathbf {0.1}, 0.2\}\)

Intercept Standard Deviation

\(\sigma _b\)

\(\{\mathbf {0.05}\}\)

Slope Mean

\(\mu _m\)

\(\{\mathbf {0.005}\}\)

Slope Standard Deviation

\(\sigma _m\)

\(\{0.001, \mathbf {0.005}, 0.01\}\)

  1. Since the range of the standard deviation depends on the range of the mean, only one parameter is varied for both the slope m and the intercept b