The article proposes a new technique for automating the screening of radiation diagnostics of employees of enterprises using elements to support medical decision-making, in particular, the U-shaped architecture of a convolutional neural network with a dual attention mechanism. A special feature of the architecture is the use of an attention mechanism based on "compression and excitation" blocks, which makes it possible to improve the quality and accuracy of digital medical data analysis, taking into account the features of computed tomography images.

Keywords: machine learning, convolutional neural network, computed tomography, architecture, chronic obstructive pulmonary disease