I came across this somewhat old question where the OP provides an unproved specific case of this identity for $n\in\Bbb N$,
$$S_n = \frac1n \sum_{k=1}^n \frac1{ak-1} = a \sum_{k=1}^\infty \frac1{ak-1} \cdot \frac1{a(n+k)-1}$$
Question 1: How do we prove this?
I fail to see how one can rearrange the finite sum into an infinite series; it almost resembles invoking a substitution while integrating.
$$\frac1n \left(\frac13 + \frac17 + \frac1{11} + \cdots + \frac1{4n-1}\right) = \frac13\cdot\frac1{4n+3} + \frac17\cdot\frac1{4n+7} + \frac1{11}\cdot\frac1{4n+11} + \cdots$$
Mathematica evaluates both sums in terms of the digamma function,
$$S_n = \frac{\psi\left(n+1-\frac1a\right) - \psi\left(1-\frac1a\right)}{an}$$
The easy part is expanding into partial fractions to recover this digamma form after finding the latter form of $S_n$, since
$$\psi(n) = -\gamma + \sum_{k=1}^\infty \left(\frac1k - \frac1{n+k-1}\right)$$
where $\gamma$ is the Euler-Mascheroni constant.
A very similar sum appeared while I was evaluating a definite integral; I managed to show
$$I = \int_0^1 \log\left(1+x^2\right) \log \left(\frac{1+x}{1-x}\right) \, \frac{dx}x = \sum_{n=1}^\infty \frac{(-1)^{n+1}}{n^2} \sum_{k=1}^n \frac1{2k-1}$$
but did not know how to continue. The identity for $S_n$ seems like a promising lead.
Question 2: Can we generalize the proof for $S_n$ and extend the identity to integer $m>1$ to rewrite
$$S_{n,m} = \frac1{n^m} \sum_{k=1}^n \frac1{ak-1} \stackrel{?}= \sum_{k=1}^\infty \frac1{ak-1} \cdot \boxed{f(a,m,n,k)}$$
It turns out that $I = \pi G - \frac74\zeta(3)$, so I think Q2's answer is affirmative in some cases.