A Hierarchical Feature Engineering Framework for Automated Classification of Phonotraumatic and Non-Phonotraumatic Vocal Hyperfunction
Researchers have developed a hierarchical feature engineering framework to automatically classify vocal hyperfunction subtypes using neck-surface acceleration data. This method integrates static, dynamic, ratio-based, and coupling features to distinguish between phonotraumatic (PVH), non-phonotraumatic (NPVH), and healthy vocal patterns. The framework achieved an AUC of 0.891 for PVH and 0.728 for NPVH, highlighting the importance of coupling features for accurate classification. AI
IMPACT Introduces a novel AI-driven approach for medical diagnosis in speech pathology.