Taking Pingshan District of Shenzhen as an example, this study started from the relationship between supply and demand, combined with mobile phone signaling and point of interest (POI) data, and used the Gaussian two-step mobile search method and p-median model to explore the accessibility and spatial optimization strategies of three types of medical facilities: general hospitals, community health centers and community clinics. The results showed that: 1) the overall spatial distribution of medical facilities in Pingshan District is “unbalanced”. The current medical facilities are mainly located in the central and northern areas of Pingshan District, and there are few in the south. 2) There is a mismatch between the accessibility of medical resources and the distribution of demand. General hospitals and community health centers are mostly distributed in the central area of Pingshan District, and the overall accessibility shows the distribution characteristics of Pingshan Street as the center and decreasing to the outer circle. Community clinics are concentrated in densely populated areas and have uneven accessibility. 3) The areas of high urgency are mostly distributed in Shatian, Tianxin, Biling, Longtian, Jiangling and Zhukeng communities, and the shortage of medical facilities in high‑urgency areas is more severe. 4) Based on the p-median model, the optimal location of three types of medical facilities is determined. The number of optimized grids for general hospitals, community health centers, and community clinics is 30, 32, and 66, respectively, with urgent needs reduced by 71.43%, 55.17%, and 97.06%. After optimization, the three types of medical facilities are evenly distributed across the district.