Unmasking Disparities in Caesarean Section Utilization in Nigeria: A K-Means Cluster Analysis of Obstetric Risk Profiles Using Machine Learning Approach

Caesarean section rates in Nigeria remain suboptimal, with significant disparities across socioeconomic and geographic strata. The objective of this research is to identify and characterize distinct obstetric risk profiles associated with caesarean section utilization in Nigeria using K-means cluster analysis, and to examine the sociodemographic and geographic factors driving these disparities. We analyzed data from 13,915 women with recent births in the 2024 Nigeria Demographic and Health Survey. Fourteen variables spanning demographics, socioeconomic status, healthcare access, medical history, and geography were used as clustering features. K-Means clustering was performed with optimal cluster selection based on silhouette score, Davies Bouldin index, and Calinski Harabasz index. Bootstrap validation with 100 iterations assessed cluster stability, while chi-square tests and logistic regression examined associations between cluster membership and surgical delivery. Ten distinct clusters were identified with rates ranging from 1.7% to 14.4%, representing an 8.4-fold variation. The highest utilization cluster at 14.4% comprised urban, highly educated, wealthy women with extensive antenatal care averaging 16.5 visits, while the lowest utilization cluster at 1.7% consisted of rural, poorly educated, impoverished women with minimal healthcare access averaging 2.3 visits. Cluster membership was significantly associated with utilization, and bootstrap analysis confirmed cluster stability with a mean silhouette of 0.220. Machine learning based clustering reveals profound disparities in utilization across distinct population subgroups in Nigeria, highlighting the dual challenge of underutilization among disadvantaged rural populations and potential overutilization among urban elites. Targeted interventions addressing geographic, economic, and healthcare access barriers are essential to optimize utilization across all population segments.