Artificial Intelligence- Driven Analysis for Pharmaceutical Product Pricing and Composition

Abstract

Increased costs in the health sector have placed significant pressure on public budgets allocated for pharmaceutical procurement. In the context of financial crises and global health emergencies, national purchasing authorities face a critical challenge of obtaining high-quality pharmaceuticals at minimal cost. Although existing literature has explored various influencing factors using producer and reference price data, limited attention has been given to individual-level purchase data from diverse public buyers. To fill this gap, this study suggests an ensemble-based AdaBoost classification method using the DrugBank database which contains about 4900 pharmaceutical compounds. The dataset is subjected to robust preprocessing, such as handling missing values, removing duplicates, implementing One-Hot Encoding, normalizing data using the Z-score, and balancing the classes using RandomOverSampler, ensuring the quality and robustness of the data for modeling. To perform comparative analysis the following multiple machine learning models are evaluated including KNN, SVM, CNN, XGBoost and the proposed AdaBoost Classifier. Experimental results show that the proposed AdaBoost classifier has excellent predictive ability and can achieve an accuracy of 99%, a precision of 99%, a recall of 99%, and an F1 score of 99%. The results show the effectiveness of ensemble learning for achieving relationships in pharmaceutical data. The study concludes that the proposed model can be used as a solution with very high accuracy and reliability in predicting the price and composition of pharmaceutical products.