It is true that Horvath and others have shown that clock deltas (between predicted and chronological age) correlate with all-cause mortality. There has been surprisingly little work on clock deltas and specific disease risk.
But these are correlational or observational studies. What is not known is whether, or how often, an intervention that changes the clock will also change lifespan or healthspan. There are far too many grants and papers that just assume this is true. Usually the argument is something like this: intervention X improves health, intervention X lowers clock age, therefore the clock is a good biomarker of aging.
The classical Horvath clock has something like 350 loci all spread out in the genome, often in regions that have no obvious functional significance.
Imagine the extreme case (which has been the subject of more than one grant proposal): take one of the clock loci 100 kbp from the nearest gene, forcibly alter its methylation state specifically. Will this change clock age? By definition, yes. Will it alter health at all? Very likely not. People seem unaware of the almost certain fact that these loci are not causal in aging, they are readouts of upstream biological states.
Also, there are numerous inconvenient facts, such as the fact that the if you compare groups of the same age, one group with and one without various neurological diseases such as Alzheimer's, the group with neurological disease appear to be younger by the clock.
I would beware of this area, there is so much hype around it and claims have far outstripped evidence by this point.