Hierarchical Classification via Cascading Feature Elimination: Application to Human Phenotype Ontology-Aligned Facial Phenotyping (FaceMesh2HPO)
The episode discusses FaceMesh2HPO, a system using hierarchical classification and cascading feature elimination to analyze 3D facial meshes. Trained on 124 clinicians' data, the model achieved a mean AUROC of 0.750. The hosts conclude that this method provides a structured, detailed profile of traits aligned with the Human Phenotype Ontology (HPO) for clinical use.