Skip to main content
Fig. 2 | BMC Neuroscience

Fig. 2

From: Objective hearing threshold identification from auditory brainstem response measurements using supervised and self-supervised approaches

Fig. 2

Scheme illustrating the two-stage method. A Example ABR plot for one stimulus. Stacked curves correspond to evoked response time signal (1000 time steps) for increasing sound pressure levels (SPL). The arrow indicates the human-assigned hearing threshold. B For each curve of all sound pressure level (SPL), a trained neural network (Model I) classifier predicts if a response is present (1, blue example) or not (0, green example), using a 1000 time step input vector and delivering a class score as output. C Result of Model I classifier. For each SPL, the binary decision (0/1) and a class score is generated. D A second classifier (Model II) uses the class score outputs from Model I as input vector and predicts the hearing threshold (HT, red). Both models are trained on the actual hearing threshold label

Back to article page