Urban environmental vulnerability to lung cancer in greater Cairo: a GIS-based multi-indicator assessment for sustainable health planning

Urban environmental, physical, residential, and socioeconomic conditions may shape contextual vulnerability relevant to lung-cancer prevention, but composite GIS maps are often interpreted more strongly than their data support. We developed a relative spatial vulnerability assessment for Greater Cairo and examined individual-level correlates of lung-cancer case status in a hospital-based case-control sample. Twenty-nine spatial criteria were organized into socioeconomic, residential, environmental, and physical domains. Criterion layers were oriented so that larger standardized values represented greater relative vulnerability, reclassified to an ordinal 1–8 scale, weighted using an analyst-defined Analytic Hierarchy Process specification, and combined by weighted linear combination under four domain-priority scenarios and one equal-weight scenario. Weight uncertainty was examined in 1,000 Monte Carlo perturbations. The complementary case-control analysis included 511 lung-cancer cases and 256 hospital-based controls recruited during 2022–2024 from a single institution, the National Cancer Institute, Cairo University, which is a national oncology referral center serving patients from governorates across Egypt and diverse socioeconomic backgrounds. Controls comprised patients with malignancies other than lung cancer and visitors or accompanying persons with no known lung-cancer diagnosis. Questionnaire categories followed a prespecified codebook, and residence codes were used only as nominal keys for an exact district-score lookup. A complete ten-predictor logistic model and an internally cross-validated radial-basis-function support vector machine were reported. Neither analysis was treated as independent validation of the GIS index. All specified AHP matrices for which a consistency ratio was defined met the prespecified CR < 0.10 criterion (maximum CR = 0.0216). Under equal-domain weighting, criterion-weight ranks were stable to ± 15% perturbations (mean Spearman ρ = 0.9912); this quantifies weight-rank stability rather than spatial or outcome validation. The maps showed higher relative vulnerability in parts of western, northwestern, and southern Greater Cairo, with patterns varying by scenario. Environmental-domain classes covered 118.087 km² (low), 626.825 km² (medium), and 660.491 km² (high); physical-domain classes covered 140.073, 635.396, and 629.935 km², respectively. In the hospital-based sample, smoking, proximity to pollution sources, and the residential-area vulnerability score had the largest positive coefficients. The logistic model had an AUC of 0.82 (95% CI 0.78–0.86); the SVM had internally cross-validated accuracy of 79%, sensitivity of 81%, and specificity of 76%. The GIS outputs identify relative vulnerability under explicit weighting assumptions; they do not estimate lung-cancer incidence or individual probability. The participant models provide complementary associations within a referral-hospital sample, not external validation. The framework is most appropriately used for environmental surveillance, hypothesis generation, and prioritization of future population-based validation.