Artificial intelligence (AI) is poised to transform clinical allergy practice by enhancing diagnostic accuracy, personalising treatment, and streamlining healthcare delivery. This narrative review critically examines the current landscape of AI in allergy care, spanning clinical workflows, diagnostics, immunotherapy, and research applications. AI-powered tools such as clinical decision support systems (CDSS), natural language processing (NLP), and conversational agents are being integrated into allergy services, offering improvements in documentation, risk stratification, and remote patient engagement—particularly in paediatric and multilingual settings. Diagnostic innovations include machine learning models that predict oral food challenge outcomes and interpret multi-omics data for personalised allergy phenotyping. AI also supports adaptive immunotherapy dosing, remote monitoring via wearable biosensors, and digital coaching to promote adherence. Federated learning and explainable AI (XAI) emerge as pivotal developments—enabling privacy-preserving collaboration and fostering trust among clinicians and patients. Despite these advancements, significant challenges remain. These include data inequities, algorithmic bias, lack of real-world validation, and regulatory ambiguity. The “black box” nature of many models risks undermining clinician confidence, while over-reliance on alerts could contribute to alarm fatigue. Ethical concerns—particularly around transparency, consent, and liability—require urgent attention. Equitable implementation demands robust governance, diverse training data, and inclusive design that prioritises patient safety. Looking ahead, AI has the potential to power digital twins, support augmented reality training, and enhance allergy surveillance through the integration of environmental and population-level data. With multidisciplinary collaboration, transparent oversight, and patient-centred innovation, AI can help build a more predictive, efficient, and equitable future for allergy care.
Qiyu et al. (Thu,) studied this question.