Drug Recommendation System for Cancer Patients Using XAI: A Traceability Perspective

Implementing Artificial Intelligence (AI) in cancer drug recommendations holds promise for advancing personalized cancer therapy. However, a key challenge faced by current AI-based drug recommendations is their lack of transparency, which hinders understanding and trust among doctors and patients. Explainable Artificial Intelligence (XAI) is a research field dedicated to designing AI systems that provide transparent explanations for their decisions. XAI-addresses the black-box nature of many AI models, aiding in humans' interpretation of internal workings and decision-making processes. This paper presents a modular approach that combines XAI techniques with a drug recommendation system based on cancer omics data. The primary objective of this approach is to offer transparent and interpretable drug recommendations specifically tailored to precision oncology. By leveraging the traceability perspective, our proposed methodology enhances the explainability of drug recommendations, thereby improving their accuracy, reliability, and trustworthiness. By incorporating XAI techniques, this research aims to bridge the gap between AI-based drug recommendations and the understanding of clinicians and patients. The traceability rate, a metric indicating the proportion of recommendations accompanied by explainable justifications, achieved a rate of 59.24%. A total of 70,211 drug recommendation predictions were made, with associated probabilities available for 41,593 predictions. The results of this study demonstrate the successful integration of cancer omics data and XAI techniques, effectively enhancing the transparency and interpretability of AI-driven drug recommendations in cancer research. This advancement contributes significantly to precision oncology by enabling informed decision-making and fostering trust in AI-based drug recommendations.