RadMaps: A Geospatial Framework for Simultaneously Modelling Capacity and Geographic Constraints on Radiotherapy Access Journal: Nuclear Physics B
Laurence Wroe, Archie Brown, Sophia Martin, Alika Ho, Steinar Stapnes · 2026
Read RadMaps: A Geospatial Framework for Simultaneously Modelling Capacity and Geographic Constraints on Radiotherapy Access Journal: Nuclear Physics B by Laurence Wroe, Archie Brown, Sophia Martin, Alika Ho, Steinar Stapnes (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-23456.txt.
Introduction
Wroe and colleagues build RadMaps, an open mapping framework that models radiotherapy access through both machine capacity and travel distance at once, where earlier studies handled each factor alone. Running on Uber's H3 hexagonal grid, it merges people counts, cancer rates, and clinic lists. Demand then goes greedily nearest-first under both limits. Applied worldwide with a 200-kilometer threshold, capacity alone allows 70 percent access and geography alone 91 percent, but together only 60 percent, exposing combined gaps planners can target.
Catalog record
| CREATOR | Laurence Wroe, Archie Brown, Sophia Martin, Alika Ho, Steinar Stapnes |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 35788 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.23456
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
RadMaps: A Geospatial Framework for Simultaneously Modelling Capacity and Geographic Constraints on Radiotherapy Access Journal: Nuclear Physics B Laurence Wroe, Archie Brown, Sophia Martin, Alika Ho, Steinar Stapnes Abstract Background: Access to radiotherapy is constrained by two compounding factors: insufficient machine capacity to meet patient demand and geographic distance from treatment facilities. Existing analyses address these factors separately, constraining the insights available to planners and policymakers. This paper presents RadMaps, an open-source geospatial framework that simultaneously models capacity and geographic constraints on radiotherapy access at any spatial scale. Methods: RadMaps operates on Uber’s H3 hexagonal grid and integrates population density data with national cancer incidence estimates and radiotherapy facility inventories. Radiotherapy demand is estimated using cancer-site-specific radiotherapy utilisation rates, and geographic access is modelled via configurable decay functions (uniform, step, or Weibull) using either distance, driving time, or public transport time, with travel times obtained via the TravelTime API. A greedy nearest-first allocation algorithm assigns demand to facilities subject to both capacity and geographic constraints, producing a localised access metric for every H3 hexagon. Results: Applied globally at H3 Resolution-3 with a 200 km 200\text{\,}\mathrm{k}\mathrm{m} step-function threshold, RadMaps computes a capacity-only access of 70 % 70\text{\,}\% , a geography-only access of 91 % 91\text{\,}\% , and a combined radiotherapy access of 60 % 60\text{\,}\% , illustrating the compounding effect of capacity and geographic constraints to significantly reduce effective access. High-resolution analyses of six countries (Australia, India, Nigeria, Oman, the United Kingdom, and the United States) demonstrate the tool’s ability to localise access deficits at sub-national scale and reveal distinct access profiles: capacity-limited (United Kingdom), geographically-limited (Oman), and doubly-constrained (India and Nigeria). Conclusions: RadMaps provides a flexible, open-access framework for visualising and identifying radiotherapy access gaps at regional to global scales, with applications in infrastructure planning and policy prioritisation. RadMaps’ modular framework is also readily extensible to other spatial access modelling applications. 1 Background 1.1 Introduction Cancer is a leading cause of mortality worldwide, with an estimated 20 million new cases and 9.7 million deaths in 2022, projected to rise to approximately 35 million annual cases by 2050 ( 3 ) . One of the principal treatment modalities for cancer is radiotherapy, which is required by approximately half of all patients for curative, adjuvant, or palliative purposes ( 6 ; 2 ) . There is, however, a well-recognised global shortfall of access to radiotherapy, driven by two compounding constraints. The first is a capacity constraint . The majority of radiotherapy is delivered via external beam radiotherapy (EBRT), in which a megavoltage machine (MVM) — predominantly a linear accelerator (linac) — delivers a targeted beam of ionising radiation to damage cancer cells ( 11 ) . Operating these machines is resource-intensive: linacs are expensive, require stable local infrastructure, dedicated shielding, and ongoing maintenance, and their clinical operation requires a multidisciplinary team of oncologists, medical physicists, dosimetrists, nurses, and engineers ( 7 ) . Using a commonly accepted benchmark of treating 450 patients per machine per year ( 22 ) , the global fleet of approximately 17,000 machines falls well short of demand. This represents an immediate shortfall of roughly 7,000 based on current case volumes, with long-term projections suggesting that an additional 30,470 linacs will be required by 2045 to keep pace with growing global demand ( 14 ) . The second is a geographic constraint . Treatment is typically delivered in daily fractions (individual treatment sessions) over several weeks. This can place a strain on populations living far from a facility, who may therefore receive suboptimal treatment or none at all ( 15 ) . A recent geospatial analysis found that 76\text{\,}\% of the global population lives within 2\text{\,}\mathrm{h} of travel time to a radiotherapy facility, with this figure dropping to 17\text{\,}\% in low-income countries ( 20 ) . In practice, these two constraints interact. Capacity-based analyses, which compare national demand with total machine counts, provide a valuable picture of the theoretical shortfall but do not capture the reduced access faced by populations far from any facility. Geographic-based analyses, such as that of Wawrzuta et al. ( 20 ) , illustrate the proximity to and distribution of facilities but do not account for the finite capacity of individual machines. Modelling both constraints simultaneously offers a more complete picture of radiotherapy access. To our knowledge, no existing tool simultaneously models both constraints on a continuous multi-scale, multi-national spatial framework. This paper presents RadMaps, an open-source geospatial tool that simultaneously models capacity and geographic constraints on radiotherapy access at any spatial scale, from sub-national to global. The tool operates on a hexagonal spatial grid and combines high-resolution population density data with national site-specific cancer incidence estimates, radiotherapy facility inventories and locations, and configurable geographic access decay functions. We describe the methodology, evaluate it through demonstration across a range of geographic and socioeconomic contexts, and discuss limitations and directions for future work. 2 Methods 2.1 Spatial Framework RadMaps operates on H3, a hierarchical geospatial indexing system developed by Uber that partitions the globe into hexagonal cells at 16 discrete resolutions ( 19 ) . H3 offers several advantages over administrative boundaries or rectangular grids for spatial accessibility analysis including: each hexagonal cell has an approximately equal distance from its centroid to all six neighbouring centroids, reducing directional bias; cell areas remain near-uniform across latitudes, with the ratio of largest to smallest cell area below 2:1 at all resolutions; and the hierarchical structure allows efficient aggregation across resolutions. RadMaps supports resolutions from Resolution-3 (average cell area of 12\,400\text{\,}\mathrm{k}\mathrm{m}^{2} ) down to Resolution-7 ( 5.16\text{\,}\mathrm{k}\mathrm{m}^{2} ). 2.2 Data Sources RadMaps uses three primary datasets. Population. Population density data are taken from the Kontur Global Population Density dataset ( 13 ) , which provides population estimates at H3 Resolution-8, integrating data from multiple sources covering the period March 2020 to November 2023. Population data are aggregated into parent hexagons at the chosen resolution in line with the H3 hierarchy. Cancer Incidence. Cancer incidence estimates for 36 different cancer types across 185 countries are taken from GLOBOCAN 2022 ( 3 ) . Radiotherapy Facilities. Facility-level data were obtained from the IAEA DIRAC database, which includes centre names, country, and inventories ( 12 ) . The database utilised in RadMaps contains 8,592 radiotherapy facilities, 8,471 of which operate at least 1 MVM, for a total of 17,097 MVMs. For the purposes of modelling, all MVMs are treated as equivalent linacs with a default capacity of 450 patients per machine per year. Geographic coordinates were added to each facility by geocoding addresses against OpenStreetMap and Google Maps. Travel Time. RadMaps supports geographic access modelling based on distance, driving time, or public transport time. While distance is computed directly within the H3 framework, driving and public transport times are integrated via TravelTime’s H3 API ( 16 ) , using typical traffic conditions departing at 08:00 on a Wednesday. 2.3 Radiotherapy Demand Estimation The radiotherapy demand is calculated as D=\sum_{i}\lambda_{i}I_{i}, (1) where I_{i} is the incidence for cancer site i and \lambda_{i} is the radiotherapy utilisation (RTU) rate for the cancer site. The RTU is specified using one of three approaches: 1. Site-specific optimal RTU: Cancer incidence per site is multiplied by the corresponding optimal radiotherapy utilisation rates as defined by Delaney