Evaluating Child Survival Predictors in Nigeria Using the Cox Proportional Hazard and Gaussian Accelerated Failure Time Models

This study assessed estimated child survival rates in Nigeria using data extracted from the 2018 Nigeria Demographic and Health Survey (NDHS). The main objectives is to stimulate and model child survival using semi parametric (Cox proportional hazards) and parametric (Gaussian accelerated failure time) survival models and to compare their clinical trails. letter survival. The study explores the influence of various factors, including geographic region, maternal education, household wealth, sanitation, water access, and gender, on child survival. Descriptive statistics reveal that the average age of first birth is approximately 19 years, with a mean age till death of 107 months. Key findings show that both models takes into account identify s factors such as regional disparities, maternal education, sanitation processes, and socioeconomic status as determinants of child survival. Children r in the Northeast and Northwest are at higher risk of death while children in the East, South-South, and Southwest regions are at lower risk of death. Additionally, lower parental education and economic hardship were linked to elevated child mortality rates, while higher education and wealth contributed to improved survival rates. Inadequate sanitation practices, such as using pit latrines and relying on water from trucks or wells, were associated with an increased risk of infant mortality. Gender differences were also observed, with males facing higher mortality risks than the female. Sanitation practices, including the use of pit toilets and reliance on tanker/cart or well water, increased the risk of infant death. To assess the models’ evaluations, the study compared the Cox PH model with the Gaussian AFT model using three criteria such as the likelihood ratio (-2LL), Akaike Information Criterion (AIC), and Bayesian Information Criterion (BIC). Results showed that the Gaussian AFT model (with -2LL of 641387.8, AIC of 641431.8, and BIC of 641646.5) outperformed the Cox PH model in terms of goodness-of-fit and overall model quality.