Table 3.

Performance of the predictive models

	Training						Validation
	Sensitivity	Specificity	Accuracy	AUC	95% CI of AUC		Sensitivity	Specificity	Accuracy	AUC	95% CI of AUC
					Lower	Upper					Lower	Upper
Initial DBN a	0.786	0.754	0.756	0.836	0.794	0.879	0.842	0.594	0.619	0.772	0.653	0.891
Logistic b	0.907	0.798	0.807	0.915	0.884	0.946	0.880	0.831	0.835	0.914	0.879	0.944
DBN‐logistic c	1.000	0.994	0.995	1.000	1.000	1.000	0.947	0.123	0.212	0.559	0.486	0.633
DBN including d
Age+Orange	0.786	0.754	0.756	0.845	0.803	0.886	0.737	0.712	0.714	0.730	0.606	0.854
Age+Leather	0.875	0.708	0.720	0.853	0.815	0.890	0.737	0.700	0.704	0.751	0.623	0.880
Age+Cinnamon	0.839	0.726	0.735	0.856	0.818	0.893	0.895	0.600	0.630	0.779	0.672	0.886
Age+Peppermint	0.821	0.738	0.744	0.864	0.825	0.903	0.737	0.676	0.683	0.736	0.608	0.863
Age+Banana	0.893	0.674	0.690	0.852	0.812	0.892	0.737	0.759	0.757	0.771	0.650	0.893
Age+Lemon	0.839	0.711	0.720	0.846	0.807	0.885	0.737	0.688	0.693	0.753	0.636	0.870
Age+Liquorice	0.839	0.717	0.726	0.851	0.813	0.889	0.684	0.735	0.730	0.766	0.649	0.883
Age+Coffee	0.839	0.738	0.745	0.850	0.809	0.891	0.842	0.594	0.619	0.766	0.645	0.887
Age+Cloves	0.839	0.707	0.716	0.843	0.802	0.884	0.842	0.635	0.656	0.773	0.655	0.891
Age+Pineapple	0.786	0.754	0.756	0.844	0.803	0.885	0.737	0.729	0.730	0.750	0.637	0.864
Age+Rose	0.893	0.677	0.693	0.854	0.815	0.893	0.737	0.718	0.720	0.767	0.643	0.890
Age+Fish	0.857	0.704	0.715	0.853	0.816	0.891	0.789	0.671	0.683	0.772	0.660	0.885
Age+MMSE+Orange	1.000	0.808	0.822	0.947	0.929	0.964	0.722	0.888	0.872	0.826	0.715	0.937
Age+MMSE+Leather	1.000	0.839	0.851	0.953	0.937	0.968	0.684	0.873	0.854	0.783	0.664	0.902
Age+MMSE+Cinnamon	1.000	0.833	0.846	0.952	0.936	0.968	0.778	0.877	0.867	0.838	0.731	0.946
Age+MMSE+Peppermint	1.000	0.805	0.819	0.951	0.934	0.968	0.722	0.865	0.851	0.817	0.703	0.932
Age+MMSE+Banana	1.000	0.853	0.864	0.958	0.944	0.972	0.632	0.862	0.839	0.743	0.623	0.863
Age+MMSE+Lemon	1.000	0.840	0.852	0.951	0.935	0.967	0.684	0.837	0.822	0.767	0.652	0.881
Age+MMSE+Liquorice	1.000	0.843	0.855	0.951	0.935	0.967	0.684	0.844	0.828	0.774	0.657	0.890
Age+MMSE+Coffee	1.000	0.789	0.805	0.949	0.931	0.967	0.789	0.756	0.760	0.793	0.683	0.903
Age+MMSE+Cloves	1.000	0.833	0.846	0.956	0.940	0.971	0.737	0.830	0.821	0.792	0.679	0.905
Age+MMSE+Pineapple	1.000	0.832	0.844	0.950	0.934	0.967	0.789	0.825	0.822	0.836	0.730	0.942
Age+MMSE+Rose	1.000	0.848	0.859	0.952	0.937	0.968	0.667	0.845	0.827	0.775	0.649	0.901
Age+MMSE+Fish	1.000	0.849	0.860	0.959	0.945	0.973	0.632	0.904	0.876	0.777	0.655	0.899

^aIncident dementia was only dependent on age.

^bMultivariable logistic regression model using the variables based on the bidirectional stepwise selection.

^cIncident dementia was dependent on the variables used in the multivariable logistic regression.

^dDiscrete Bayesian networks (DBN) including dependency of incident dementia on the variables below.