Application of an AI-supported diagnostic planning tool in ENT patients

Introduction The development of AI (artificial intelligence)-based assistance systems is still in the early stages of development in the field of medicine and applications in everyday clinical practice have not yet been established. Primarily the focus is on image recognition or diagnosis assistance systems. However, the development of an instrument for planning diagnostics or therapies would be just as important for optimizing the organization of patient flows, especially against the background of scarce human resources.

Method On the basis of an automated AI-supported digital questionnaire, the patients’ symptoms were recorded and corresponding suggestions for additional diagnostic procedures were generated based on a Bayesian network. These were compared to the diagnostic procedures actually carried out.

Results The system was used in n = 19 patients in our ENT outpatient clinic. Only in 5 patients the suggested diagnostics were in accordance with the methods actually carried out, in the other 14 patients the suggestions were not congruent.

Discussion Diagnostic assistance systems are still in development. In the system we use, the focus is on suggesting a diagnosis and not the planning of diagnostics, which is why the data from the underlying Bayesian network is not yet sufficient for a specific application. Nevertheless, it seems sensible to also address this possibility of using AI in medicine, since there is considerable potential for optimization of resources.

Publication History

Article published online:
12 May 2023