Severity Prediction of Hair Loss in Alopecia Areata With Hybrid Deep Learning Architecture

The study introduces a hybrid deep learning model for classifying the severity of hair loss in alopecia areata using a dataset of 11,880 images. The model distinguishes five harm degrees with high accuracy, achieving an overall accuracy of 97.89%. Harm Degrees 1, 2, and 3 showed exceptional performance with precision, recall, and F1-scores around 98%, and class-specific accuracy of 0.99. Harm Degrees 4 and 5 also performed well with 97.81% across all metrics. These results highlight the model's potential for effective and rapid automated harm detection and severity classification in clinical settings.