Can I find fraud or unusual patterns in Medical history and/or concomitant medication?
May 18, 2017 12:04 PM(3283 views)
Does anyone have experience analysing medical history for fraud/unusual patterns. I mean finding subjects that have the same (or partly the same) medical history and concomitant medications? The duplicate records review does not allow me to choose the MH or CM domains.
You could try running the Cluster Subjects Within Study Sites Report, where you uncheck "Findings" on the options and check "Events, Interventions" instead. While there is not a filter to only use CM and MH, you could get just these domains by Adding the Study to JMP Clinical (again) and only selecting the DM, CM, MH domains for inclusion in the study analysis.
Doing this with the example Nicardipine shows the sites that have unusually low between-subject distances. In that data example, Site 40 has the minimum subject distance (has at least one set of subjects that are pretty similar based on MH and CM events).
You can select those subjects that cluster more closely together and use the Show Subjects Action to get the table of incidence counts for each MH/CM.
The JMP Cell Plot works nicely to then visualize how closely the subjects really were (aka how many events they both experienced).
Choose all the event columns as Y, Reponse and be sure to check the option to Scale Uniformly