regexp in spark sql to extract numeric from a string

I have a string

norethindrone (norethindrone 5 mg oral tablet)

from which I need to match 5

I have used the below query in oracle and got output as “5”

select regexp_replace(regexp_SUBSTR(lower('norethindrone (norethindrone 5 mg oral tablet)')
,'[0-9,./]+ ?(mc?g|ml|meq|(intl )?unit.?s?.?|g|each|dose|%)+([-0-9/ ]+(ml|mc?g))?'),'[a-z% ]+','')
from dual

But the same code in spark SQL output null.

What would be the reason?

Note: The string I gave here is a sample and I have millions of records with different possible combinations.