Validating administrative data to identify complex surgical site infections following cardiac implantable electronic device implantation: a comparison of traditional methods and machine learning

Table 2 Sensitivity, specificity, positive predictive value and negative predictive value for each approach

	Area Under Curve (ROC)	Sensitivity	Specificity	Negative predictive value	Positive predictive value
Algorithms (Traditional pre-selected codes)
Algorithm 1	0.894 ± 0.04	0.833 ± 0.08	0.955 ± 0.005	0.999 ± 0.001	0.124 ± 0.016
Algorithm 2	0.946 ± 0.064	0.906 ± 0.128	0.986 ± 0.001	0.999 ± 0.001	0.325 ± 0.049
Algorithm 3	0.816 ± 0.13	0.669 ± 0.261	0.962 ± 0.004	0.997 ± 0.002	0.115 ± 0.04
Machine learning models
ICD Model	0.961 ± 0.06	0.903 ± 0.129	0.987 ± 0.009	0.999 ± 0.001	0.378 ± 0.133
ICD & CCI Model	0.968 ± 0.043	.9 ± 0.13	0.99 ± 0.004	0.999 ± 0.001	0.416 ± 0.047
Unique Code
T827	0.936 ± 0.006	0.881 ± 0.009	0.991 ± 0.005	0.999 ± 0.0	0.428 ± 0.119

ISSN: 2047-2994