Replication data and code for "Beyond topological robustness: Structural–functional divergence between road-network connectivity and healthcare accessibility under random and targeted disruptions"

Research overview This repository provides processed road-network data, analysis code, and numerical results supporting the research article entitled "Beyond topological robustness: Structural–functional divergence between road-network connectivity and healthcare accessibility under random and targeted disruptions". The study examines the relationship between structural road-network robustness and geographical healthcare accessibility under random and edge-betweenness-centrality (EBC)-targeted disruptions in Jiangsu Province, China. Dataset contents The deposited package includes a reconstructed road network comprising 1,258,974 road-segment records and 927,810 nodes, approximate edge-betweenness centrality estimates, a fixed targeted-removal ranking, numerical results for structural and healthcare-accessibility indicators, selected sensitivity-analysis outputs, Python scripts, and supporting documentation. The principal performance measures are largest connected component (LCC), healthcare-connected node ratio (HCNR), and the proportions of road-network nodes within 5 km (HCR5) and 10 km (HCR10) of the nearest healthcare facility along the road network. The main experiments consider removal fractions ranging from 0% to 30%. Reproducibility Structural LCC results under targeted disruption can be independently recomputed from the deposited road-network data at all seven removal levels. The included verification script checks the recomputed LCC values against the corresponding reference results. Healthcare accessibility results, including HCNR, HCR5, and HCR10, are provided in numerical reference tables. However, the facility-to-network attachment data required to independently recompute these functional indicators are not included in this public release. Consequently, the functional results can be inspected and compared with the supplied reference outputs but cannot be independently regenerated from the deposited files alone. The archive also contains supporting results and scripts associated with sensitivity analyses. Some analyses depend on input data that are not redistributed, and their complete independent reproduction is therefore outside the scope of this public package. Excluded data and limitations Raw Amap point-of-interest records, facility-to-network attachment tables, facility-level connector screening results, National Healthcare Security Administration registry extracts, and the Tianditu administrative boundary layer are excluded from the public archive because the authors have not established the necessary rights to redistribute these third-party or derived spatial materials. The original ArcGIS network reconstruction was carried out through documented GIS processing operations and is not fully automated in the deposited scripts. Computational reproduction starts from the deposited reconstructed road network. Licensing and documentation Different components of the package are subject to different licenses. OpenStreetMap-derived road-network databases are distributed under ODbL 1.0, original Python code under the MIT License, and eligible research results under CC BY 4.0. The package includes README.md, LICENSE_DATA.md, LICENSE_CODE.txt, and MANIFEST_sha256.csv for documentation of the dataset contents, licensing conditions, limitations, and file integrity. Version 1.1.0 — Conservative public release excluding facility-level spatial attachment data.