Try using IBP on I_(n-1) where u=1/(x^2+1)^(n-1) and dv=1. (Haven’t done the qn but it seems like it may work). Note that doing IBP on I_n will most likely give you an (x^2+1)^(n+1) on the denominator. With this question it looks tempting to do tan inverse but it doesn’t appear in the reduced form.