BFTS: Thompson Sampling with Bayesian Additive Regression Trees

Researchers developed a new AI algorithm, BFTS, that improves personalized recommendations in apps like health interventions. It combines a proven exploration strategy with a powerful, probabilistic tree-based model to better predict complex user behavior. In tests, it achieved state-of-the-art performance on benchmarks and, in a real-world evaluation of a drinking reduction app, improved user engagement rates by over 30% compared to the previously deployed system.