The authors propose an algorithm which can aid the estimation of the validity of combined risk models lacking primary data.