Improving spatial access to healthcare facilities is of great interest in urban planning. However, most facility location models fall short of explicitly incorporating the spatial accessibility measure in designing facility locations. Thus, this paper develops a novel framework to directly improve spatial access to healthcare facilities, by integrating spatial analysis with an optimization model. Specifically, by leveraging multiple types of data, this framework could: (1) estimate healthcare demands using machine learning models from large-scale mobile phone data; (2) calibrate the travel time decay effect based on human mobility patterns and road networks; (3) calculate spatial accessibility scores; and (4) optimize the location of hospitals with the consideration of efficiency and equity criteria. To demonstrate the applicability of the proposed framework, we used Erie County in the State of New York as a case study. This study could help urban planners determine the optimal locations to add new hospitals, with the hope of improving equitable access to healthcare services.