XAI approach to drug dosage optimization and review-based prediction of side effects

One of the major advancements in medicine is “personalized medicine”: a new approach with a promising future that takes into account the unique makeup of each individual patient instead of broad populations. One of the challenges of personalized medicine is the complexity: the more variables to account for, the more complex the adaptation of a certain treatment to a certain patient becomes; thus, the need for artificial intelligence: a tool that can be used to account for the various variables, and perform the computational and mathematical analyses required to succeed in the task of personalizing a certain treatment modality. This study explores explainable artificial intelligence (XAI) for drug dosage optimization based on individual patient data, utilizing advanced machine learning techniques: Decision Trees and Random Forests to be specific. Our approach applies XAI using ML models to provide clear and interpretable dosage recommendations, as well as predicting potential side effects. The methodology proves a high adaptability to diverse patient profiles, achieving an accuracy of 70% in dosage optimization. These results showcase the efficacy of XAI and the utilized algorithms in improving personalized treatment modalities, providing clinicians with a clearer vision, and understandable dosage recommendations for tailored treatment plans.