Predicting Drug Dosage with Machine Learning: An Empirical Study Using Ceftaroline

Accurate drug dosage prediction is crucial to optimize treatment efficacy and minimize side effects, especially in vulnerable pediatric populations. This study evaluates two machine learning approaches, artificial neural networks and decision trees, to predict the initial dose of Ceftaroline in pediatric patients using clinical and pharmacokinetic data, including plasma drug concentrations. Results demonstrate that both models effectively estimate the required dose; however, the decision tree model achieves superior accuracy, with a mean absolute error (MAE) of 6.99 mg compared to 13.59 mg for the neural network. These findings highlight the potential of machine learning in improving personalized drug dosing and clinical decision-making in pediatric pharmacology.