AI is only as good as it's teaching data and learning...

AI is only as good as it's teaching data and learning algorithms. My company is using AI for audio based analysis in industry and we have to have very specifically defined and clearly labelled data for any target features to get high confidence levels. AI can be real time for automated processes but without narrowing down the process state and requisite model, the data processing requirements quickly become prohibitive.

The 4D fluid flow analysis has been developed using engineering mathematical models and validated in the real world. Physics is the underlying truth from which the 4D product creates rapid visibility of the dynamic airflow behaviour in a patient for experienced clinicians to quickly understand. It allows much faster and more specific preliminary identification of problems. The data captured is richer and more dynamically sensitive. Compared to other diagnostic methods it is lower cost for the level of information provided. It also can reduce the need for invasive tests by ruling out or in lung issues and disease location in the lung.

It may be that in time AI can be used to provide additional evaluation of the flow data but final diagnosis will still be the call of a clinician.