GSPICE is a data-driven Gaussian Process-like approach to solve the issue of spectral artefacts. Unlike traditional template-based methods, GSPICE relies solely on the data to detect artefacts and predicts the clean spectra. To do this, it uses conditional estimation where the value of a given pixel is predicted based on all the other pixels.
This package is currently under active developement.