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

The study presents a hybrid deep learning model combining CNN and Logistic Regression to classify the severity of hair loss in Alopecia Areata using a dataset of 11,880 images. The model distinguishes five harm degrees based on hair loss percentage, achieving high classification accuracy across all metrics, with an overall accuracy of 97.89%. Harm Degrees 1, 2, and 3 showed particularly high performance, each with around 98% accuracy. This model offers a reliable and rapid tool for automated harm detection, potentially improving clinical diagnosis by reducing reliance on subjective visual inspections.