Before using AI in a clinical workflow, an allergy practice should look for evidence that the tool has been tested prospectively in real-world settings. It should also plan for clinical oversight and the resources needed to put the tool into practice.
Check for real-world testing
A review of AI in clinical allergy practice reports that few allergy AI models have been prospectively validated in real-world settings. That makes the evidence for the specific tool and intended workflow an important part of a practice's review.
Owners can record what prospective testing is available and the clinical workflow the practice is considering. That gives the practice a concrete basis for deciding whether the evidence is relevant to its planned use.
Keep clinicians involved
Van Breugel, speaking on the CSACI podcast The Allergist, says clinicians who use AI have a responsibility to remain critical. A practice can make that responsibility concrete by identifying who will review the tool's outputs in the workflow and how that review will happen.
An owner's practical takeaway is to document the intended workflow, the clinician responsible for review, and the evidence considered before introducing the tool.
Account for implementation needs
The clinical allergy review identifies infrastructure and financial constraints as barriers to AI implementation, particularly in resource-poor settings. It also says effective deployment requires training and change management.
Those needs affect whether a tool can fit the practice's operations. Include the required infrastructure, expected costs, staff training and workflow changes in the practice's planning.
This page is general education only. It is not medical advice. Talk with your own doctor about your health.
Frequently asked questions
What should a practice ask a developer about validation?
Van Breugel says developers have a responsibility to validate tools thoroughly and apply proper governance. A practice can ask what validation has been completed and what governance applies to the tool.
What does limited real-world validation leave uncertain?
The clinical allergy review says the shortage of prospective real-world validation raises questions about generalisability and robustness. Practice leaders can include those questions when considering evidence for their intended setting.
Should owners account for future regulation?
Van Breugel noted that more regulation is arriving. Her comment does not specify particular requirements, so practices can include regulatory developments in their ongoing oversight planning.
Sources
- Artificial intelligence in clinical allergy practice: current status, challenges, and future directions (Exploration of Asthma & Allergy)
- The Allergist, Episode 69 transcript: When AI Meets the Allergy Clinic (The Allergist podcast (CSACI))
Immunologists.com is a business directory and education site. It does not give medical advice, rank or rate care, or refer patients. Listings come from public registry data and are not endorsements.
