Skip to main content
. 2018 Aug 22;8:12597. doi: 10.1038/s41598-018-30722-0

Table 1.

Predictive log likelihood ratio for each model and subject.

Model Subject
CV GB LB PG SM SE TA VR VD
LNE 1.1678 (0.5012) −1.8267 (0.2976) −0.3094 (0.2020) 0.1516 (0.3407) 0.4896 (0.3427) −1.7151 (0.3493) −0.0365 (0.2806) 0.3456 (0.3593) 0.3793 (0.2175)
EKF 1.1900 (0.1998) 0.0380 (0.1326) 0.7168 (0.2234) 0.3828 (0.1428) −1.4262 (0.2755) 1.0909 (0.1543) 0.2358 (0.1250) 0.3623 (0.1581) 0.6537 (0.1002)
BLSmem 0.4621 (0.1832) −0.0821 (0.0267) −0.2445 (0.0743) 0.0086 (0.0518) −0.0637 (0.2714) 0.0440 (0.0876) −0.0191 (0.0195) −0.0359 (0.0249) −0.0421 (0.0268)

For each subject, the model with the highest predictive log likelihood is shown in bold. All log likelihood ratios were all smaller than 2 and consistent with what is expected based on simulations of each model (Supplementary Figs 7, 9 and 11).