|
Abstract:
|
We develop a probability model to schedule the next cancer screening exam dynamically based on an individual's screening history. That is, when to schedule the next screening exam for an asymptomatic individual with or without a history of screenings? We first derive the conditional probability of incidence as a function of the next screening time interval, given an asymptomatic woman with or without a screening history, then we can find the screening time interval by limiting this probability to a small value, such as 5% or 10%. That is, with 95% or 90% probability one will not become a clinical incident case before the next scheduled exam. This conditional probability of disease incidence changes with one's age, screening history, screening sensitivity, sojourn time in the preclinical state, and transition probability into the preclinical state, etc. We will use the Health Insurance Plan for Great New Yorker (HIP) breast cancer screening data to illustrate this new method.
|
