Skip to main content
. 2021 May 18;9(5):1223–1239. doi: 10.1007/s43390-021-00360-0

Fig. 5.

Fig. 5

Using a predictive model to simulate patient MCID and complication rates. A total of 1245 patients were simulated using gradient boosting classification, to determine predicted MCID and complication rates. Incremental thresholds were then applied at 10% intervals, to identify sub-populations of patients meeting specific outcomes criteria. The region highlighted by the red box represents patients with predicted MCID > 50%, and complication rate < 20%. This group corresponded to 33% of the original cohort and using these criteria, cost reduction information and decreased surgical utilization rates were extrapolated for ASD surgery in the United States and in Spain