Not to be confused with a computing cluster, this is a reference to analytical methods.
In medicine and medical jargon, the verb “to cluster” and noun “cluster” have a specialized definition. It’s a loaded term, intended to signal that observations of coincidental medical conditions found close together and at around the same time are presumed to be strongly related in an as yet unknown way.
It’s usually stated in a public health context, when an alert or warning is announced.
Example: A sudden spike in food poisoning at restaurants is noticed by way of shared reporting between hospitals. If there are 10,000 incidents within 48 hours in the entire state of Texas, statistical “clustering” helps shake out the unrelated noise of the usual incidental food poisoning, to locate the unexpected events according to shared qualities. Same restaurant chain, correlated ingredients across chains (all lettuce), associated factories or farms, suppliers, freight shipping company, style of prep work, common equipment (all had same blender or mixer brand), and so on.
It’s a way of problem solving during fluid events as they unfold, in an attempt to get out in front of a problem, as it happens.