Where should we build a new health facility? Where the largest population can be served

At SDSC NYC 2025, Dr. Joan LaRovere from the Virtue Foundation shared how geospatial data and optimization models are used to answer a critical question: where should a new health facility be built to save the most lives? Their model identifies the best locations by analyzing multiple data layers to minimize travel time and serve the largest populations, ensuring facilities aren't built on impractical locations like mountains or lakes. This analysis, conducted across 72 countries, has revealed surprising insights, such as the need for healthcare access in the suburbs of major capitals in low and middle-income countries. Inspired by this clip? Watch the full talk and all sessions from SDSC NYC 2025 on demand: https://spatial-data-science-conference.com/2025/newyork#Watch-on-demand #SDSC25 #geospatial #datascience #healthcare #VirtueFoundation Transcript: 00:02 Regarding the optimization model and the work our data science team has done with DataRobot, we took our foundational data, coupled it with other data layers, and trained the model. The model was instructed not to place a hospital on a road, mountain, or lake, but where the largest population can be served and travel time can be diminished. 00:24 We then asked the model to rank the top 20, 10, 5, and 1 locations. This allows for decisions to be made beyond government policy makers who shout the loudest. You can now geospatially determine where to place a new facility. We have also applied this to existing facilities. 00:46 This helps answer questions like, "Where should we open the next cardiac service or ophthalmology service?" One insight we found, while looking at Ghana and we have run this across all 72 countries, is the presence of clusters. The suburbs are the most desirable locations. While we might think of rural areas, it is the suburbs of major capital hubs in low and middle-income countries that need more access to care. 01:16 These findings are unique to this method of analysis. This is the raster we created, and we've run this across all 72 countries for travel time to facilities. The dark red areas indicate you are far from a facility, while the whiter areas indicate you are close. We then couple that with population and facility data. 01:41 The dark green areas indicate better coverage, and the white to red areas indicate poorer coverage. Due to time constraints, I cannot show you this for specialties and subspecialties, but we have done this at each level.