Here's a great explanation of SSTables, which include Statistics.db and summary.db files. SSTables are created when a new table space is created. The format of the SSTable is embedded in the file name. In this example, the format is version "ka".
main_dev-table1-ka-56-Statistics.db
From the reference:
In version "ka", contents of Statistics component file is split in to
3 types of metadata called validation, compaction and stats.
Validation metadata is used to validate SSTable which includes
partitioner and bloom filter fp chance fields. Compaction metadata
includes ancestors information which is also available in older
formats and a new field called cardinality estimator. Cardinality
estimator is used to efficiently pre-allocate bloom filter space in a
merged compaction file by estimating how much the input SStables
overlap. Stats metadata contains rest of the information available in
old formats and two additional fields. First one is a flag to track
the presence of local/remote counter shards and the other one is for
storing the repair time.