HealthCheck: abnormally huge number of calls to an interface

We have this defect showing up as "High" in the HealthCheck (done each 2w) of our Appian infra:

• [Process Model Name] (44.4% of all CPU time, 67.6% of all operations, 117 mins of CPU time, 3213596 operation count)

We don't have a huge number of processes of this type per day (around 600 complete), and the number of nodes is low. We also see in the same HealthCheck that the input interface it uses is called 1 million times!

• Interface.[Process Model Input Interface] (1056180 calls, 113 ms peak average)

There is no way our one hundred users could call the interface one million times - there would be many more processes per day.

1. Has anyone ever encountered such a repeated call to an input interface, causing huge CPU usage for a single process?
2. Is it possible that the 1'500 such "Active" processes (kept as drafts of user input) be somehow regularly polled, thus calling the input task, and cause the huge number of calls to the interface ?
3. If the above is possible, how do we change the polling frequency, or change the PM to wait on a node that do not call the interface and crush my CPU?

Here is why I am making the above hypothesis (2):