Fig. 10From: Sparse multi-output Gaussian processes for online medical time series predictionThe percent improvement using MedGP for online imputation compared to independent (unvariate) GPs. The figures depicts the results of 24 covariates for the (a) sepsis, (b) heart failure, and (c) neoplasms and (d) MIMIC-III heart failure subgroups. The y-axis is on log scale. The error bars denote ±1 standard error. The ⋆ indicates statistical significance p < 4.17 x 1e-4Back to article page