Overview
Paul S. Albert is a biostatistician at the Biometric Research Branch, National Cancer Institute (NCI). His work focuses on latent class models for diagnostic accuracy, longitudinal and correlated data methods, and statistical methods for cancer research. He is distinct from James H. Albert (Bowling Green State University), who works on Bayesian computation for categorical data.
Key Contributions / Features
- Albert-Dodd (2004): Demonstrated that latent class models for estimating diagnostic accuracy without a gold standard can produce widely divergent sensitivity estimates even when fitting equally well; established conditions (J ≥ 10) for reliable model identification. See Albert-Dodd (2004).
- Albert et al. (2001): Extended latent class models for diagnostic accuracy with a finite mixture approach allowing a proportion of subjects to be always correctly classified.
Related