Fig. 7.

Latent state model for cervical cancer screening data. The matrix $\mathbf{M}\in {\mathbb{R}}^{N\times T}$ of latent state profiles decomposed into female-specific coefficients $\mathbf{U}\in {\mathbb{R}}^{N\times r}$ and a small number ($r\ll min\left\{N,,,T\right\}$) of basic profiles $\mathbf{V}\in {\mathbb{R}}^{T\times r}$