Fitting Plackett-Burman design on child disease experiment

Abstract

A child health experiment was designed to achieve the maximum positive variables through the screening of different disease variables, using the Plackett-Burman design. Eleven variables of the children disease were identified: Acute Bronchial Meningitis, Respiratory tract infection, chronic liver disease, Congestive cardiac failure, upper respiratory tract infection, Urinary tract infection, Ante-partum Haemorrhage, Post-partum Haemorrhage, Sickle cell disease, Neonatal Jandice, and Benigu prostal hyperplasia. The selected variables were evaluated through statistical analysis, based on their significance, coefficient value and standard effect plot. The results suggested that three variables, namely, upper respiratory tract infection, urinary tract infection, Ante-Partum Haemorrhage and the combination of urinary tract infection/Ante-Partum Haemorrhage had influence with high confidence levels, while the remaining eight variables did not show significant effect on the children age. The coefficient of determination value R2 (63.42%) also showed the model used for prediction to be significant (p less than 0.05). The plot for the standard effect for each component and its traits provided accurate data by which to select well-suited variables and further statistical optimization of medium and process parameters was explored using stepwise regression methodology.