Dementia prediction in the general population using clinically accessible variables: a proof-of-concept study using machine learning. The AGES-Reykjavik study

Table 3 Summary of the elastic net models on test data (n = 1870), as well as stratified by sex

	AUC	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)
Model 1	0.73 [0.70–0.76]	61 [55–66]	71 [69–73]	33 [29–37]	89 [87–91]
Women	0.74 [0.70–0.78]	60 [53–67]	70 [67–73]	35 [30–40]	87 [84–89]
Men	0.73 [0.67–0.79]	64 [55–73]	71 [67–75]	29 [23–35]	92 [89–94]
Model 2	0.73 [0.70–0.76]	61 [56–66]	71 [69–74]	33 [29–37]	89 [87–91]
Women	0.73 [0.69–0.77]	59 [52–66]	71 [67–74]	35 [30–40]	87 [84–89]
Men	0.73 [0.67–0.79]	63 [54–72]	72 [68–76]	29 [24–36]	92 [89–94]
Model 3	0.72 [0.69–0.75]	61 [56–66]	69 [66–71]	31 [28–35]	89 [86–90]
Women	0.71 [0.67–0.75]	59 [52–65]	69 [66–72]	33 [29–38]	86 [84–89]
Men	0.72 [0.66–0.78]	66 [57–75]	67 [63–71]	27 [22–33]	92 [89–94]

AUC = area under the ROC curve; PPV = positive predictive value; NPV = negative predictive value

ISSN: 1472-6947