I used the excellent GPT 5.6 Sol to create a library for time series analysis in Fortran by translating 48 R packages and plan to add more from R and other languages. If you download the source and documentation for an R package from GitHub, Sol can quickly translate it to modern Fortran. It can also compare the capabilities of the package against what already exists in the library and only translate distinct algorithms. Licenses of the R package are preserved. Coding algorithms has become almost free, and the main tasks now are verifying and organizing them. There is a program for automatic model selection for univariate time series that loops over many of the implemented models. As @certik suggested, we can translate SciPy back to Fortran.
Related projects are