Predict Method for JointODE Objects

Computes predicted biomarker trajectories, velocities, and accelerations for subjects based on the fitted joint ODE model. Predictions incorporate both fixed effects and subject-specific random effects.

Usage

# S3 method for class 'JointODE'
predict(
  object,
  newdata = NULL,
  times = NULL,
  parallel = FALSE,
  n_cores = 0,
  ...
)

Arguments

object

An object of class JointODE

newdata

Optional data frame with new subjects. If NULL, uses the training data from the model fit.

times

Optional time points for prediction. Can be:

• A numeric vector for the same times across all subjects

• A named list with subject-specific time vectors

• NULL to use observation times from the data (default)

parallel

Logical flag for parallel computation (default: FALSE)

n_cores

Number of cores for parallel processing. If 0, automatically detects available cores (default: 0)

...

Additional arguments (currently unused)

Value

A data.frame with columns:

id

Subject identifier

time

Time points for predictions

cumhaz

Predicted cumulative hazard at each time point

survival

Predicted survival probability, computed as \(S(t) = \exp(-\text{cumhaz})\)

log_hazard

Log instantaneous hazard at each time point

biomarker

Predicted biomarker values

velocity

Predicted velocity (first derivative)

acceleration

Predicted acceleration (second derivative)