Deep learning‐based prediction of treatment prognosis from nasal polyp histology slides

The study aimed to develop a prognostic indicator for patient outcomes by analyzing scanned conventional hematoxylin and eosin (H&E)-stained slides of nasal polyps using deep learning. An interpretable supervised deep learning model was developed and validated on an internal cohort of 185 H&E-stained whole-slide images (WSIs) of nasal polyps and an external cohort of 122 H&E-stained WSIs. A poor prognosis score (PPS) was established and applied to visualize the histopathological features underlying PPS. The model yielded a patient-level sensitivity of 79.5% and specificity of 92.3% on the external cohort, with areas under the receiver operating characteristic curve of 0.943. The predictive ability of PPS was superior to that of conventional tissue eosinophil number. The study concludes that the deep learning model is an effective method for decoding pathological images of nasal polyps, providing a valuable solution for disease prognosis prediction and precise patient treatment.