This work presents an extension from previous formalism on period search in gamma-ray AGN light-curves, consisting of the inclusion of a new time-series approach able to analyse irregularly sampled data. It consists of the inclusion of a Continuous Auto-Regressive first-order formalism CAR(1) to describe the stochastic coloured noise present in the AGN emission. With this, the available number of sources to study is increased significantly. These light-curves are fitted
using a Markov Chain Monte-Carlo minimization, being able to separate the stochastic and the deterministic components. Here, the new algorithm is applied to a sub-sample of AGN candidates from the Fermi-LAT catalogue. Finally, the results can be reviewed to better understand physical models explaining periodic high-energy AGN behaviour.