Research Paper Volume 12, Issue 8 pp 7561—7575

Using a genetic algorithm to derive a highly predictive and context-specific frailty index

Figure 4. (A) Distribution of the number of selected deficits for the best FIs in each iteration among the 10 genetic algorithm cycles; (B) mean AUC and 95% confidence intervals in the prediction of 3-year and 6-year mortality in the whole population and in sex- and age subgroups (calculated in the complete dataset) for 100 randomly generated FIs including 40 deficits (mean AUC for ga-FI, calculated in the complete dataset, shown in red); (C) mean AUC in the prediction of 3-year and 6-year mortality in the whole population and in sex- and age subgroups (calculated in the complete dataset) for more than 2000 randomly generated FIs including 25-108 deficits (mean AUC for ga-FI, calculated in the complete dataset, shown in red) – boxplots show median and 2nd and 3rd quartiles of mean AUC for FIs with similar number of deficits.