Fig. 2: Incorporating decoder outputs in the MINDFUL score maintains a high correlation with performance. | Communications Biology

Fig. 2: Incorporating decoder outputs in the MINDFUL score maintains a high correlation with performance.

From: Measuring instability in chronic human intracortical neural recordings towards stable, long-term brain-computer interfaces

Fig. 2

a The KLD (right y-axis) between distributions of decoded directional vectors, \(\hat{X}\), with respect to the sub-selected time bins from the first session(s) overlaid onto median AE (left y-axis in blue) across all recorded sessions for T11 (left panel) and T5 (right panel). Subsequent neural distributions and median AE were updated every 1 s over a 60-s sliding window. Pearson r, and Spearman rank correlation coefficients ρ, between KLD and median AE are shown. b The KLD between distributions of \(\hat{X}\) and \({\hat{X}}_{{lag}}\) overlaid onto median AE. c The KLD between distributions of the combination of derived neural features (as shown in Fig. 1c), decoded directional vectors, \(\hat{X}\), and \({\hat{X}}_{{lag}}\), overlaid onto median AE.

Back to article page