How to Use AI to analyse test execution output (SQL output) to design a regression suite?

We currently run SQL reports to extract test execution output so that we can review how successful a test has been and then make an educated guess of which tests to add to our regression suites.

However this is time consuming as it requires someone to go through all the data and make certain assumptions.

I've been tasked with looking into the possibility of using artificial intelligence to sift through the data instead and would like to know if anyone has tried this and how they implemented.