Within the framework of General Relativity, is there a well-defined way of determining how the curvature affects the group velocity of a wave packet, more specifically a localised gravitational wave? I know there is such a thing as gravitational lensing of waves, but what I'm aiming for is an equation (similar to the geodesic equation?) that somehow relates the group velocity to the connection or the curvature. More specifically I'm talking about the case where the metric of the space is of the form: $$\gamma_{\mu \nu} \approx g_{\mu \nu} +h_{\mu \nu}$$ where $g$ is the metric of the curved background, and $h$ describes the wave packet and is sufficiently small. So the intended equation would probably involve only first-order terms in $h$.
When I searched online for answers, I found some papers on accelerated wave packets (i.e their centre of mass velocity, or group velocity changes with time) but they only treated particular cases, for example, a wave packet in a spherical shell, and the acceleration is only given explicitly.
I suppose defining such an equation may be problematic for several reasons: the group velocity is usually defined w.r.t to an absolute time coordinate, and I don't know how one would go about defining a "group 4-velocity". Also, due to dispersion, the equation may not always hold, and if it does it will only approximately hold for the short spacetime interval where the wave packet is localised. If such an equation exists, can it be derived from the wave equation in curved space-time (derived from the sourceless field equations)? How would one go about defining such as an equation?
Any help or further clarification on the subject would be greatly appreciated!