I'm interested in a simplification for the integral given by $$\int_{\lambda}^\infty \int_{0}^\infty e^{-(s+t)}(st)^{x-1}\ln \frac{s}{t} ds dt$$
which arises as a definition for the density of a random variable following the "continuous Poisson" distribution.
This isn't a complete answer.
Here is a way to convert the double integral into two simple integrals (that are numerically more tractable).
Write the initial integral
$$\int_{\lambda}^\infty \int_{0}^\infty e^{-(s+t)}(st)^{x-1}\ln \frac{s}{t} ds dt$$
into
$$\int_{\lambda}^\infty e^{-s}s^{x-1}\int_{0}^\infty e^{-t}t^{x-1}(\ln s - \ln t) dt ds$$
$$=\int_{\lambda}^\infty e^{-s}s^{x-1}\ln s \underbrace{\left(\int_{0}^\infty e^{-t}t^{x-1}dt\right)}_{\Gamma(x) \ \text{independent of} s} ds - \int_{\lambda}^\infty e^{-s}s^{x-1}\underbrace{\left(\int_{0}^\infty e^{-t}t^{x-1}\ln t dt\right)}_{I=\Gamma^{\ \prime}(x)} ds$$
For the proof of $I=\Gamma'(x)$, see for example here.
Sorry to stop there, wishing you further advances by different techniques (integration by parts, etc...)