Prognostic Gamma-Power Generalized Regression Modelling of Determinants Influencing Variations in Under-Five Mortality Rate

The under-five mortality rate is a serious health indicator in sub-Saharan Africa which has negative right skewed heteroscedastic nature that restricts the conventional regression models. This hypothesized the Gamma-Power-Log-Logistic (GPLL) prognostic framework to estimate determinants of under-five mortality rate in Lagos State, Nigeria, with the aid of primary data of 476 women of childbearing age (15-49 years) in five administrative divisions in 2025. The GPLL model incorporates a power-transformed distribution into a generalized linear model through a theoretically motivated transformation process, which makes it possible to perform likelihood-based inference that can deal with skewness, heavy tails and bounded support. Findings were that the new GPLL regression is significantly better than the gaussian linear and Gamma regression benchmarks, with maternal education and access to healthcare proved to be important protective factors and larger household size leads to higher risk of mortality. The framework effectively models the spatial heterogeneity and tail behaviour not in the other traditional models. The paper is concluded to be based on the evidence of flexible distributional modelling by showing reliable assessment of mortality rates and suggest specific intervention points that should focus on female education and access to health services as well as family planning to lower the under-five mortality rates in urban African environments. This study directly addresses the United Nations Sustainable Development Goal (SDG) 3.2 addressing preventable deaths of children under five years of age.