somewhat skeptical.

as I understand it, by substituting the 'wrong' disease during the mathematical process phase (why not just subsitute a secret code?) the mathematical process can be done without revealing things. Probably true, but for research to be done, at some point that must be stripped away. Diseases are not abstract, researchers must know the diseases they are dealing with to make any sense of the data.

Additionally it is essential for other researchers to have access to the raw data to validate the judgements and analysis techniques used. This analysis is not clean. There is always a big risk of confounding factors that the original researcher missed, and without exact knowledge of how conditions were classified, which ones were removed from the dataset and why, it is impossible to spot those confounding factors.