Uncloaking the black-box: the need for explainable artificial intelligence in clinical microbiology and infectious diseases applications

Antimicrobial resistance and emerging infectious diseases remain significant challenges for global health, driving a need for advanced technological solutions. Artificial Intelligence (AI) expanded opportunities in clinical microbiology, infectious diseases, and public health by harnessing vast, structured datasets. Despite impressive analytical capabilities, the clinical integration of AI-based applications is hindered by its opacity. The "black-box" aspect undermines adoption into healthcare workflows. Explainable AI (XAI) methods, including intrinsically interpretable models and post-hoc interpretability tools, such as SHAP, LIME, and Grad-CAM, can address these transparency challenges. This narrative review is intended to be a primer for the interested clinician. It systematically evaluates recent advancements in XAI in the context of clinical applications for clinical microbiology, infectious diseases, and public health. We further discuss the ethical and regulatory landscape shaping AI adoption, including the critical role of open, quality-controlled data, robust performance metrics, and clear interpretability to ensure safe and effective clinical implementation. Lastly, we propose future directions, emphasizing interdisciplinary collaboration, international data-sharing initiatives, and tailored AI literacy training to facilitate trustworthy, equitable, and impactful use of AI in clinical microbiology and infectious diseases.