"ghost" EC2 on Autoscaling Groups

We are running into some weird issues without our EC2 and Autoscaling groups. Sometimes one of our instances starts failing checks, and we have to terminate it. Once we do that, the instance looks to be terminated from the EC2 dashboard and API, but somehow the autoscaling group associated with it thinks the instance still exists. When we go into the "Instance management" tab in the autoscaling group you can see the instance id of the "ghost" instance is still there. If you click on it to see details it opens a page with a loading spinner and "It is taking a bit longer than usual to fetch your data", which never completes. This is not temporary: that instance was terminated days ago. Because the autoscaling group still thinks the instances exist it never spins up a new one to meet the minimum instance requirements. The autoscaling activities doesn't show any logs or events related (ie.: trying to terminate instance or start a new one, it just thinks everything is working normally).

Our EC2 dashboard showing a single instance on the region:

The instances that the autoscaling group thinks it's associated with. The first one is the same as the one above, and the one selected is the one that "doesn't exist"

Instance detail page once you click on it

Selecting the instance and clicking "detach" solves the issue, but this has happened a few times and we would like to get to the root of the issue.