A MIXTURE MODEL FOR INTERVAL-CENSORED TIME-TO-RESPONSE QUANTAL ASSAY DATA

Various mixture models are proposed and fitted to a set of time-to-response quantal assay data. The models provide a natural extension to the time-dependent case of the classical threshold model for quantal assay data. The approach deals simply with interval censoring, and provides flexible families of distributions, for both the mixing probability and the cumulative distribution function for time to response. For data of the kind considered in this paper, mixture models may provide a more suitable description than other models from survival analysis, such as the proportional hazards model.