Table 1.

Model performance for each patient and model, displayed via balanced accuracy values and the sensitivity for “Math” and for “Non-Math” (in brackets, respectively). Those values represent averages over the 10 folds.

MKL model	Patient P1	Patient P2	Patient P3
Full	89.51 (88.06–90.96)	87.18 (85.49–88.89)	76.80 (72.89–80.71)
δ	73.48 (69.25–77.71)	82.25 (81.18–83.33)	72.36 (69.01–75.71)
θ	70.92 (68.66–73.16)	79.89 (78.43–81.35)	73.08 (69.37–76.79)
α	75.28 (71.64–78.92)	75.57 (71.37–79.76)	75.03 (70.77–79.29)
β	66.88 (64.18–69.58)	78.50 (78.04–78.97)	69.18 (65.14–73.21)
low-γ	85.61 (84.18–87.05)	81.07 (79.61–82.54)	68.81 (66.20–67.57)
high-γ	88.92 (86.27–86.86)	88.37 (87.45–89.29)	69.50 (69.37–69.64)