There are just two tables/models based on experimental data? What about Haldane, Hawkins, Yarbrough, Van der Aue, Buhlmann, BSAC, DCIEM, NAUI, and VPM? Do they not count? Or are they not commonly used? Or were they not considered? It would appear that all of these have various degrees of experimental grounding. There seems to be less that one case of DCS per thousand dives which seems to be a success rate well outside of voodoo territory.
It is hard to know if a given computer is safe since the manufacturers typically do not share the details of what they are doing. That said it appears what is commonly done is to use one of the more modern models listed above as points of departure, and then the algorithm is typically made more conservative. How well that works out is hard to say since there is not generally data. Also it is possible to go too far the other way. Some manufacturers have a penalty for cold water. While in principle that may make sense the dive computer does not have insight into how well the divers dry suit is working on a given day or even if a dry suit is being used. So the implementation at times is dubious.