I don't know what the state of the art is and if this directly addresses your question, but perhaps it would be interesting to conduct an exploratory analysis: having a dataset of mapped dolines (polygons) of known (relative) age or evolutionary stage and their DTMs, one could compute a range of metrics to run through a statistical analysis aimed at identifying which morphometric parameters correlate with assumed relative evolutionary stage. Ideally, one should have as much control as possible over morphometry-conditioning factors, i.e., use a sample of dolines from a relatively homogeneous area.