heisenberg-meth/Bayesian-Survival-Models ? reverse-engineered prompt

Build me a Python research project for survival analysis that compares four approaches on a few clinical datasets: a standard Cox model, a random survival forest, a simple deep learning survival model, and a Bayesian Cox model. I want it to download and prepare the data, train each model, and then compare them with survival metrics like concordance and Brier score across cross validation.

Please make it easy to run each experiment from the command line, and save the results, charts, and summary tables into a reports folder. I also want basic statistical checks so I can tell whether one model is really better than another, plus a few explainability outputs if possible.

Keep the whole thing reproducible, with fixed seeds and config files for the main settings. If you need to, look up the current docs for the libraries involved and make the setup work cleanly in a fresh Python environment.