The authors propose a novel Bayseian nonparametric regression approach to the construction of synthetic controls, enabling the supplementation of experimental data with non-experimental data.