Validation of Machine Learning-Based Assessment of Major Depressive Disorder from Paralinguistic Speech Characteristics in Routine Care

Table 5

Machine learning classifications based on depression scale items.

	Balanced accuracy mean (SD)
	Exclusive model	Fused model
HRSD-based	0.72 (0.07)	0.73 (0.07)
QIDS-C-based	0.75 (0.07)	0.76 (0.07)
PHQ-9-based	0.73 (0.07)	0.73 (0.06)

The exclusive models only use scale items, whereas the fused models combine audio features and scale items. HRSD = Hamilton Rating Scale for Depression; QIDS-C = Quick Inventory of Depressive Symptomatology–Clinician-rated; PHQ-9 = depression module of the Patient Health Questionnaire.