By: Fatima Khan, MD, FACAAI, Chair of AI and Technology Subcommittee
The past few months have marked a turning point in the evolution of health‑focused artificial intelligence. Major technology leaders – OpenAI, Anthropic’s Claude, and Amazon’s One Medical have each launched new clinical‑facing AI services designed to support documentation, patient communication, triage, and care navigation. For the first time, large‑scale AI platforms are being built explicitly for health care delivery rather than adapted from consumer tools. This shift signals the beginning of a new era – one that will reshape how clinicians practice, how patients engage with care, and how specialty fields like allergy and immunology evolve.
A particularly notable development is the rise of consumer‑connected health AI, where patients can directly link their personal health data: labs, wearables, medication records, and more to AI systems that generate individualized insights. These tools promise to help patients better understand their health trends, interpret test results, and prepare questions for their clinicians. For many specialties, this may enhance engagement and self‑management. For allergists, the implications are both exciting and complex.
However, this trend introduces unique challenges. Much of our diagnostic work relies on tests that require contextual, nuanced interpretation: skin testing, serum IgE levels, component testing, and provocation challenges. These results cannot be interpreted in isolation. An AI system that ingests raw data without clinical context may generate explanations that are incomplete, misleading, or simply incorrect. A patient may arrive convinced they have an allergic condition based on an elevated IgE level, or that a mild sensitization explains chronic symptoms, when the true picture is far more complex. The risk is not just misinformation; it is misplaced confidence. And increasingly, patients may generate these insights without ever seeing an allergist for proper guidance.
The rapid commercialization of health AI raises important questions. How do we ensure accuracy, safety, and equity when models are trained on heterogeneous data? What data have they been trained on that is relevant to our specialty? What guardrails are needed to prevent over‑reliance on algorithmic suggestions? How do we preserve the clinician-patient relationship in an era when digital tools may become the first point of contact? These concerns are central to how allergists maintain trust, uphold quality, and deliver patient‑centered care.
As these tools proliferate, allergists will increasingly encounter patients who bring AI‑generated summaries, risk scores, or interpretations to their visits. I have already experienced patients sharing ChatGPT‑generated diagnoses and recommendations for workup based on symptoms they entered. Rather than viewing this as a disruption, we can prepare in ways that strengthen the clinician-patient relationship so we remain the experts patients turn to first.
The College’s AI and Technology Subcommittee is already working to provide practical guidance and educational resources. Our goal is not to promote any specific platform, but to empower clinicians to navigate this landscape with confidence.
We are entering a transformative era. Health AI will not remain optional; it will become embedded in the fabric of care. By engaging early, having an open mind, asking the right questions and centering on patient well‑being, allergists can lead rather than react. Allergists cannot control which platforms patients use, but we can ask practical questions that help us contextualize the output and help our patients become more discerning users of AI and maintain our role as trusted experts. This new era offers tremendous promise if we approach it with curiosity, caution, and a commitment to excellence to ensure AI augments – not replaces – the clinical reasoning, empathy, and contextual judgment that define our specialty.
