Title: Fast Approximations for Binary and Ordinal Data in Dynamic
Systems Models

Author: Charles Driver

Abstract:
This talk explores a computationally efficient approach for incorporating
binary and ordinal measurement models into longitudinal and dynamic systems
models, using a modified extended Kalman filter. I demonstrate the Kalman
filter and the additional modifications, and compare performance to a full
Bayesian sampling approach. The approaches are implemented in the ctsem
software for R, offering a practical solution for modeling discrete or
continuous-time systems with a mixture of continuous, binary, and ordinal
response variables.