Fig. 4: Model performance in severity prediction across cognitive groups.
From: Evaluating spoken language as a biomarker for automated screening of cognitive impairment
MAE for predictions on the DementiaBank test set (N = 71 participants, each with one MMSE score). Central points represent the mean predicted MMSE across 10 bootstrap repeats for each participant. Error bars show the standard deviation across repeats as an estimate of model uncertainty.
