Clinical Evaluation of AI-aided Auscultation With Automatic Classification of Respiratory System Sounds

Crackle, Lung Sound, Rhonchi, Wheezing

In the present trial, the StethoMe electronic stethoscope will be used for pulmonary auscultation and recording of lung sounds in the pediatric study population.

Multicenter, international - 2 EU sites (max. 1 in Poland) and a single US site. Trial centers and investigators will be identified and selected based on their clinical and research experience. The trial objective is to assess whether use of the StethoMe AI lung sounds analysis software provides clinical benefit by improving the identification of abnormal lung sounds in the categories of wheezes, rhonchi, fine crackles and coarse as compared to pulmonary auscultation by experienced physicians (general practitioners (GPs) and pulmonologists). The recordings will be used to form a gold standard database of lung sounds as evaluated by an expert, independent panel according to the study protocol. The gold standard database will include approximately equal representation of wheezes, rhonchi, fine crackles and coarse crackles. Moreover, the database will also include recordings without any of the previous descripted pathological sounds. The results of these two analyses, by traditional physician listening and by the StethoMe AI software application, will be assessed for sensitivity and specificity to the gold standard database in detection of the four lung sounds and recordings without defined pathological sounds. The StethoMe AI will be considered to provide clinical benefit if it demonstrates statistically better sensitivity or specificity on any of the four lung sounds as compared to the traditional physician auscultation, together with F1 score analysis.

No related papers found yet.