2025A&A...704A.150A

Context. Transit and radial velocity (RV) techniques are the dominant methods for exoplanet detection, while astrometric exoplanet detections have been very limited thus far. Gaia has the potential to radically change this picture, enabling astrometric detections of substellar companions at scale that would allow us to complement the picture of exoplanet architectures given by transit and RV methods.
Aims. Our primary objective in this study is to enhance the current statistics of substellar companions, particularly within regions of the orbital period-mass parameter space that remain poorly constrained by RV and transit detection methods.
Methods. Using supervised learning, we trained a deep neural network (DNN) to recognise the characteristic distribution of the fit quality statistics corresponding to a Gaia Data Release 3 (DR3) astrometric solution for a non-single star. We created a deep learning model, ExoDNN, which predicts the probability of a DR3 source to host unresolved companions.
Results. Applying the predictive capability of ExoDNN to a volume-limited sample (d<100 pc) of F, G, K, and M stars from Gaia DR3, we have produced a list of 7414 candidate stars hosting companions. The stellar properties of these candidates, such as their mass and metallicity, are similar to those of the Gaia DR3 non-single-star sample. We also identified synergies with future observatories, such as PLATO, and we propose a follow-up strategy with the intention of investigating the most promising candidates among those samples.