Table 2.
Internally cross-validated performance of the machine-learning models in the pooled two-center cohort.
| Task | Different methods | AUROC (95% CI) | AUPRC(95% CI) | Brier Score (95% CI) | Accuracy (95% CI) | Recall(95%CI) | Specificity (95%CI) | NPV (95% CI) |
|---|---|---|---|---|---|---|---|---|
| Bowel Resection |
RF | 0.86 (0.82-0.90) |
0.58 (0.51-0.66) |
0.115 (0.095-0.135) |
0.78 (0.73-0.82) |
0.83 (0.75-0.91) |
0.76 (0.71-0.82) |
0.95 (0.93-0.97) |
| XGBoost | 0.85 (0.81-0.89) |
0.56 (0.48-0.65) |
0.120 (0.098-0.142) |
0.76 (0.71-0.81) |
0.83 (0.72-0.94) |
0.75 (0.68-0.82) |
0.95 (0.93-0.98) |
|
| CatBoost | 0.84 (0.78-0.90) |
0.58 (0.50-0.66) |
0.117 (0.097-0.139) |
0.74 (0.70-0.79) |
0.76 (0.65-0.87) |
0.74 (0.68-0.79) |
0.93 (0.90-0.96) |
|
| AdaBoost | 0.81 (0.75-0.87) |
0.51 (0.42-0.61) |
0.131 (0.111-0.153) |
0.71 (0.64-0.77) |
0.69 (0.51-0.86) |
0.71 (0.62-0.80) |
0.91 (0.87-0.96) |
|
| KNN | 0.79 (0.74-0.85) |
0.47 (0.39-0.56) |
0.128 (0.108-0.151) |
0.73 (0.68-0.77) |
0.69 (0.51-0.86) |
0.74 (0.66-0.81) |
0.92 (0.88-0.96) |
|
| SVM | 0.83 (0.79-0.88) |
0.62 (0.57-0.68) |
0.114 (0.094-0.136) |
0.75 (0.71-0.80) |
0.77 (0.66-0.88) |
0.75 (0.70-0.79) |
0.93 (0.90-0.97) |
|
| GBDT | 0.85 (0.81-0.89) |
0.60 (0.52-0.68) |
0.117 (0.097-0.139) |
0.76 (0.72-0.81) |
0.77 (0.64-0.90) |
0.76 (0.72-0.81) |
0.94 (0.90-0.97) |
|
| Perianal Complication |
RF | 0.71 (0.64-0.78) |
0.61 (0.53-0.69) |
0.212 (0.195-0.229) |
0.68 (0.61-0.74) |
0.64 (0.54-0.75) |
0.70 (0.65-0.75) |
0.76 (0.70-0.83) |
| XGBoost | 0.71 (0.63-0.79) |
0.63 (0.54-0.71) |
0.211 (0.191-0.233) |
0.69 (0.64-0.75) |
0.70 (0.60-0.80) |
0.69 (0.63-0.76) |
0.79 (0.74-0.85) |
|
| CatBoost | 0.68 (0.60-0.75) |
0.60 (0.52-0.68) |
0.219 (0.205-0.235) |
0.63 (0.58-0.69) |
0.66 (0.55-0.78) |
0.61 (0.54-0.69) |
0.76 (0.68-0.84) |
|
| AdaBoost | 0.70 (0.63-0.77) |
0.59 (0.52-0.65) |
0.217 (0.200-0.235) |
0.67 (0.63-0.72) |
0.72 (0.62-0.82) |
0.64 (0.60-0.69) |
0.80 (0.73-0.86) |
|
| KNN | 0.65 (0.58-0.72) |
0.57 (0.50-0.64) |
0.221 (0.206-0.238) |
0.59 (0.55-0.64) |
0.53 (0.39-0.67) |
0.64 (0.56-0.72) |
0.70 (0.63-0.76) |
|
| SVM | 0.70 (0.62-0.77) |
0.62 (0.54-0.70) |
0.210 (0.193-0.227) |
0.65 (0.60-0.69) |
0.60 (0.47-0.73) |
0.68 (0.63-0.73) |
0.75 (0.68-0.81) |
|
| GBDT | 0.69 (0.63-0.76) |
0.62 (0.55-0.70) |
0.215 (0.201-0.228) |
0.66 (0.61-0.70) |
0.64 (0.54-0.74) |
0.67 (0.62-0.71) |
0.76 (0.70-0.81) |
|
| Abdominal Complication |
RF | 0.73 (0.69-0.78) |
0.52 (0.44-0.59) |
0.160 (0.137-0.182) |
0.68 (0.63-0.73) |
0.71 (0.65-0.78) |
0.66 (0.59-0.74) |
0.89 (0.87-0.91) |
| XGBoost | 0.72 (0.67-0.76) |
0.51 (0.44-0.59) |
0.163 (0.140-0.187) |
0.63 (0.55-0.70) |
0.61 (0.52-0.70) |
0.63 (0.53-0.74) |
0.84 (0.81-0.88) |
|
| CatBoost | 0.67 (0.62-0.73) |
0.46 (0.38-0.54) |
0.166 (0.144-0.188) |
0.65 (0.59-0.72) |
0.57 (0.48-0.67) |
0.68 (0.60-0.75) |
0.84 (0.81-0.88) |
|
| AdaBoost | 0.71 (0.65-0.77) |
0.47 (0.38-0.55) |
0.165 (0.143-0.191) |
0.68 (0.61-0.75) |
0.61 (0.47-0.74) |
0.70 (0.60-0.79) |
0.86 (0.82-0.90) |
|
| KNN | 0.63 (0.58-0.68) |
0.41 (0.34-0.49) |
0.168 (0.146-0.191) |
0.62 (0.58-0.67) |
0.58 (0.50-0.67) |
0.63 (0.58-0.69) |
0.84 (0.81-0.87) |
|
| SVM | 0.67 (0.60-0.74) |
0.43 (0.34-0.52) |
0.168 (0.146-0.192) |
0.61 (0.54-0.67) |
0.58 (0.50-0.65) |
0.62 (0.53-0.70) |
0.83 (0.79-0.86) |
|
| GBDT | 0.68 (0.63-0.73) |
0.46 (0.39-0.54) |
0.166 (0.145-0.187) |
0.63 (0.55-0.71) |
0.61 (0.51-0.72) |
0.64 (0.53-0.74) |
0.84 (0.80-0.89) |
Values are presented as the mean across the 10 held-out outer folds, with descriptive fold-based 95% intervals reflecting variation across the outer-fold partitions.