Though of course this is the method that led to the discovery of Neptune: https://en.wikipedia.org/wiki/Discovery_of_Neptune
Yes. Refreshing my memory from the link you gave, it seems like the Neptune searchers would have benefited decisively from image-differencing and all-inclusive source databases -- like the ones that are now automated. Because, during their search for Neptune, they observed it twice without noticing it:
"Only after the discovery of Neptune had been announced in Paris and Berlin did it become apparent that Neptune had been observed on August 8 and August 12 but because Challis lacked an up-to-date star-map, it was not recognized as a planet."
http://crts.caltech.edu/index.html
The original survey was built for near-Earth objects, but the extra-solar-system transients detected (e.g., supernovae) became the subject of many papers.
The supernova catalog that observations like these compiled -- but from other transient-detection networks -- became the basis of the 2011 Nobel Prize for the observational discovery of dark energy: https://www.nobelprize.org/nobel_prizes/physics/laureates/20...
You may be seeking after something harder, which is to use orbit anomalies to infer positions of objects - kind of an automation of all the orbital simulation efforts described in the article. That, of course, has not been done.