Research Project
3492054
CTM, A Program for the Estimation and Testing of Continuous Time Multi- State Multi-Spell Modeling, has been funded by grants from NIH to the University of Chicago to embody recent major advances in modeling of survival and transition data. It includes modeling of several types of parametric and nonparametric unobserved heterogeneity (state and person specific), multiple spell data (such as birth spacing intervals, repeated episodes of remission from disease) and multi-state data (as in markov processes), and simulations for goodness of fit tests, all features not found in other programs. Currently this complex stand along program is extremely difficult to use, its output difficult to interpret, its successful performance often requires expert intervention, and it is implemented only on mainframes. We propose to simplify and abbreviate the program integrating CTM into the SYSTAT package, and rendering a version for micros. We would provide an easy to use, fault tolerant, error handling, user assisting interface, together with enhanced intelligence within the program to propose diagnosis of estimation problems, and automatic adjustment of estimation strategies within the program. The integration to SYSTAT will also provide connection to high resolution graphics services, effective database management and conventional statistical routines.
