Combining Deep Learning and Machine Learning for the Automatic Identification of Hip Prosthesis Failure: Development, Validation and Explainability Analysis

•Development of a combined DL-ML approach to automatically detect hip prosthetic failure.•Extraction of deep features from 3 images (total, acetabular and stem)•Accuracy of 0.958 ± 0.006 in the internal validation (cohort 1 – 280 patients)•F1-score of 0.874 in the external validation set (cohort 2 – 352 patients)