Introduction of Artificial Intelligence (AI) to Clinical Decision Support System (CDSS) has brought a new paradigm in personalized pharmacotherapy, making possible the choice and optimization of medication according to the individual characteristics of patients. This paper proposes a methodology for AI-enabled CDSS for improving therapeutic decisions based on machine learning, natural language processing, and deep learning technologies. Methodology: The framework for the CDSS includes three major components, including: multimodal data fusion component for combining genomic and clinical data; predictive analytics module using gradient-boosted decision trees and deep neural networks; and the explainable AI component using SHAP and LIME approaches. Results: The results obtained from simulations and real-life datasets indicate the effectiveness of the proposed framework, as it achieves 94-96% accuracy in performing drug recommendation, significantly improving the precision in predicting adverse events and optimizing dosages. Conclusion: It is concluded that AI-powered CDSS can improve patient safety and therapeutic outcomes, decreasing medication errors and advancing from population-based medicine towards personalized pharmacotherapy.