Latent trajectories of frailty and risk prediction models among geriatric community dwellers: an interpretable machine learning perspective

Table 1 Performance of machine learning algorithms for frailty trajectories prediction

	Accuracy	Recall	Precision	F1 Score	AUROC	Brier Score
NB	0.645 (0.610, 0.680)	0.810 (0.787, 0.832)	0.802 (0.777, 0.827)	0.805 (0.782, 0.829)	0.722 (0.680, 0.764)	0.161 (0.143, 0.179)
LR	0.664 (0.627, 0.702)	0.701 (0.674, 0.728)	0.798 (0.773, 0.823)	0.733 (0.709, 0.757)	0.728 (0.687, 0.769)	0.205 (0.196, 0.213)
DT	0.612 (0.580, 0.643)	0.819 (0.798, 0.839)	0.794 (0.769, 0.820)	0.802 (0.779, 0.826)	0.612 (0.580, 0.643)	0.208 (0.202, 0.214)
RF	0.610 (0.580, 0.639)	0.845 (0.825, 0.864)	0.820 (0.794, 0.846)	0.816 (0.791, 0.841)	0.702 (0.659, 0.745)	0.143 (0.134, 0.151)
SVM	0.596 (0.560, 0.632)	0.723 (0.697, 0.748)	0.765 (0.737, 0.793)	0.740 (0.716, 0.765)	0.661 (0.619, 0.703)	0.181 (0.167, 0.195)
XGB	0.599 (0.580, 0.639)	0.833 (0.825, 0.864)	0.802 (0.794, 0.846)	0.806 (0.791, 0.841)	0.677 (0.659, 0.745)	0.134 (0.134, 0.151)
ANN	0.595 (0.563, 0.628)	0.796 (0.773, 0.819)	0.777 (0.750, 0.804)	0.785 (0.760, 0.810)	0.640 (0.595, 0.685)	0.193 (0.171, 0.215)

Note: NB Naïve Bayes, LR Logistic regression, DT Decision tree, RF Random forest, SVM Support vector machine, XGB Extreme Gradient Boosting, ANN Artificial neural network, Accuracy refers to balanced accuracy, Precision, Recall, and F1 score refer to the weighted results.

ISSN: 1471-2318