Fig. 4: Modality weights (acoustic, linguistic, demographic) assigned by the SpeechCARE Adaptive Gating Fusion (AGF) network across various speech tasks in the test set. | npj Digital Medicine

Fig. 4: Modality weights (acoustic, linguistic, demographic) assigned by the SpeechCARE Adaptive Gating Fusion (AGF) network across various speech tasks in the test set.

From: SpeechCARE: dynamic multimodal modeling for cognitive screening in diverse linguistic and speech task contexts

Fig. 4

Each cluster of bars represents one task, showing how the model differentially weights acoustic (light blue), linguistic (dark blue), and demographic (purple) features depending on the cognitive demands of the task.

Back to article page