The broad objective is to develop statistical methods for analyzing family history data. Specifically, theoretical and computer models for analyzing age of onset data will be developed. These methods are applicable both to aggregation and segregation analysis. The methods will be applied initially to large data sets of breast and colon cancer which will be accessed by the sponsor. The models will be used to understand the genetic basis of disease and familial clustering of age to onset data. These models can be readily applied to the estimation of cancer risk for high risk individuals given their family history and environmental risk factors. Such methods are important for counseling and cancer control efforts. Computer software for the analysis will be developed and made public. Likelihood methods for the study of correlated survival times arising from a mixture distribution will be developed. Mixture models provide a flexible, powerful and interpretable framework for research in genetic epidemiology. The primary tool in analyzing such data is the EM and allied algorithms. These tools will be used to study age of onset data. The performance of these methods is to be compared with existing generalized estimating approaches and those based on exponential regression.
| Etzioni, R; Cha, R; Feuer, E J et al. (1998) Asymptomatic incidence and duration of prostate cancer. Am J Epidemiol 148:775-85 |
