Integration of lung ultrasound with clinical variables in AI models

Investigate methods to combine lung ultrasound imaging with clinical variables such as anamnesis and complete blood count within artificial intelligence models, and determine how lung ultrasound compares to these variables in predictive performance.

Background

Although lung ultrasound is one of many clinical examinations, most existing AI work focuses on image-only approaches without leveraging clinical context such as symptoms or laboratory values. This gap limits understanding of the comparative value and synergy between modalities.

The authors emphasize that integrating clinical variables with lung ultrasound within AI frameworks has not been sufficiently explored, and explicitly state that it is unclear how to perform such integration or comparison.