Machine learning to predict untreated dental caries in adolescents

BMC Oral Health

Table 2 Machine learning approach and metrics for adolescents. SBMS 2018/2019. (n = 615)

Collect metrics	xgboost		Decision trees		Logistic regression
Sensitivity	0.42(CI95% ± 0.02)		0.44(CI95% ± 0.02)		0.40(CI95% ± 0.02)
Specificity	0.92(CI95% ± 0.02)		0.88(CI95% ± 0.02)		0.80(CI95% ± 0.02)
Acuraccy	0.75(CI95% ± 0.01)		0.79(CI95% ± 0.01)		0.76(CI95% ± 0.01)
AUC	0.84 (CI95% ± 0.01)		0.81(CI95% ± 0.01)		0.73(CI95% ± 0.01)
Three Main contributors	importance*	Three Main contributors	importance*	Three Main contributors	importance*
Dental Floss**	0.37	Unhealthy**	0.20	Unhealthy**	0.15
Unhealthy consumption **	0.30	Whites	0.17	Dental Floss**	0.13
Whites	0.17	Dental Floss**	0.14	Fluoridation	0.09

* importance ranges from 0 to 1. The main gain in information, the more important the predictor role on the outcome ( untreated dental caries)
** Possibly modified factors to be addressed for Primary health care workers

ISSN: 1472-6831