Fig. 2.

(A) Beta weights of laboratory indicators that differed significantly between the two groups. (B) QS effect size of the significant indicators. Significant QS values indicate that we can reject the null hypothesis of equal distribution. QS values indicate the quantile to which the distribution has shifted from the population median (0.5). (C) The odds ratio of death by laboratory indicators: Single-index Models. “High” represented odds ratio of death by indicator values more than reference range, “Low” was similar meaning. (D) The average predictive power of indicators across 10 iterations. To be more cautious and clinically significant, subjects with a probability of fatality of more than 0.3 were deemed non-survivors in the prediction procedures.