In this document, we propose a new method that combines high-dimensional data with machine learning methods to predict physician-hospital integration. We compare the performance of this method with alter- native approaches used in the healthcare economics literature for a large validated sample, finding that it outperforms previous methods by a substantial margin. We also compare the static predictions of this model to a large validated sample recently made available by the Agency for Healthcare Research and Quality (AHRQ) and again document a high degree of accuracy. Finally, we briefly summarize the implications of our method for the growth in physician-hospital integration over the years 2008-2016.