Machine-learning based prediction of appendicitis for patients presenting with acute abdominal pain at the emergency department

Table 2 Statistical comparison of AUROC values among ML models, ED physicians, and the Alvarado scoring system using DeLong’s Test

Comparison	AUROC ± CI	P value
ML models
HIVE	0.919 ± 0.023	0.978
HIVE-LAB	0.923 ± 0.020	0.978
HIVE model / ED physicians without lab
Physician 1	0.919 ± 0.023 / 0.894 ± 0.076	0.375
Physician 2	0.919 ± 0.023 / 0.826 ± 0.106	0.037
Physician 3	0.919 ± 0.023 / 0.791 ± 0.117	0.007
HIVE-LAB model / ED physicians with lab
Physician 1	0.923 ± 0.020 / 0.923 ± 0.067	0.796
Physician 2	0.923 ± 0.020 / 0.892 ± 0.078	0.353
Physician 3	0.923 ± 0.020 / 0.859 ± 0.098	0.118
ED physicians without / with lab
Physician 1	0.894 ± 0.076 / 0.923 ± 0.067	0.182
Physician 2	0.826 ± 0.106 / 0.892 ± 0.078	0.058
Physician 3	0.791 ± 0.117 / 0.859 ± 0.098	0.177
ML models / Alvarado
HIVE vs. Alvarado	0.919 ± 0.023 / 0.824 ± 0.095	0.033
HIVE-LAB vs. Alvarado	0.923 ± 0.020 / 0.824 ± 0.095	0.031
ED Physicians without lab / Alvarado
Physician 1	0.894 ± 0.076 / 0.824 ± 0.095	0.247
Physician 2	0.826 ± 0.106 / 0.824 ± 0.095	0.980
Physician 3	0.791 ± 0.117 / 0.824 ± 0.095	0.646
ED Physicians with lab / Alvarado
Physician 1	0.923 ± 0.067 / 0.824 ± 0.095	0.071
Physician 2	0.892 ± 0.078 / 0.824 ± 0.095	0.240
Physician 3	0.859 ± 0.098 / 0.824 ± 0.095	0.599

ISSN: 1749-7922