Fig. 3From: Learning temporal weights of clinical events using variable importanceCompare weights learned on different specificity levels in terms of predictive performance. Green indicates a significant difference and red indicates no significance. P-values are obtained via a Friedman testBack to article page