Are you saying that you want to fetch the DecisionResult at a later period? Is there any specific reason for such requirement?
Here are the following options that you could evaluate:
Prior to 7.2.1 the delayed learning of ADM has following option:
Perform Make decision and store the decision results into pxDecisionResults(DDS dataset)
Retrieve the stored decision results and ADM payload during set response using the subject id and interaction id combination(these are the keys for pxDecisionResults dataset) and pass it on to strategy as external input
The output of the response strategy is used to update the ADM via pxAdaptiveAnalytics
This way of retrieving responses involved storing the make decision interaction id's explicitly and later on use it during set response.
In 7.2.1, with the introduction of Event store the need of storing interaction id after make decision can be eliminated as the make decision results are stored in pxDecisionResults as well as the meta data(interaction id) is also stored in pxEventStore.
When sending a response in 7.2.1, user can select the way in which the previous decisions for a given subject id should be fetched.
Using Interaction id : which is the older implementation where user has to send in the subject id and interaction id along with the response for fetching the previous decision.
Using Time range : which is the new implementation where user can specify a time period (in minutes, hours and days) and all the previous decisions for a subject id in that specified time period will be fetched.