Deep learning boosted amyloidosis diagnosis

Amyloid light chain (AL) amyloidosis is a disorder characterized by the deposition of antibody light chains in organs. Early and accurate diagnosis of AL amyloidosis is crucial for timely implementation of appropriate treatment strategies. However, existing computational methods for predicting AL amyloidosis often heavily rely on manually extracted features and their performance is less than satisfactory. In this study, we introduce DeepAL, a deep learning-based approach designed to predict AL amyloidosis with high precision. DeepAL utilizes a pre-trained model to extract light chain features and is then fine-tuned with AL amyloidosis knowledge. On two benchmark datasets, DeepAL achieved impressive results with area under the ROC curves (AUCs) of 0.9072 and 0.8919, outperforming previous approaches. Our ablation study shows the use of the pre-trained model can significantly improve identification performance. The code is available at https://github.com/waterlooms/DeepAL.