Given the current usage of regular expressions to match collection names in the aggregation pipeline, we should investigate whether switching to string or substring comparison might yield better performance. Regular expressions are known to be computationally intensive, especially when used repeatedly across multiple conditions.