3 ms·

I imagine getting samples/training sets that don't have some hefty source of bias is really tough in medicine, and probably leads to over-fitting.