Emergency Care Units (UPAs) are part of a national health policy implemented by the Brazilian Government. UPAs are fixed prehospital components of the Brazilian Unified Health System (SUS), whose purpose is to provide resolutive emergency care to patients suffering from acute clinical conditions, and to perform the first care in cases of surgical nature. According to the Ministry of Economy, 750 units are operational throughout the country since 2008, and 332 are under construction. Being a public policy in expansion, it is imperative to assess the impact of such units as part of SUS. However, we found few studies that assessed UPAs’ impact, which have examined their specific impact on mortality rates. In our research, we aimed to evaluate the impact of UPAs on hospitalization rates for diseases of the respiratory system. To measure the impact, we used a strategy of Machine Learning through the Bayesian Additive Regression Trees (BART) algorithm. The results point to a decrease in the hospitalization rates by respiratory diseases due to Emergency Care Units. Therefore, these units generate a benefit for the Brazilian health system, being an important element for the care of patients with respiratory diseases.