Machine learning analyses identify multi-modal frailty factors that selectively discriminate four cohorts in the Alzheimer’s disease spectrum: a COMPASS-ND study

Table 2 Random Forest Model Evaluation Metrics

RF Model	Indicators Tested	AUC	Accuracy	Precision	Recall	F₁ Score
Discriminating SCI from CU	n = 64	0.89	0.79	0.94	0.58	0.72
Discriminating MCI from CU	n = 65	0.88	0.76	0.75	0.95	0.84
Discriminating AD from CU	n = 75	0.98	0.88	0.98	0.73	0.84

Evaluation metrics reflect average performance of the RF classification model across the five cross-validation folds. Each evaluation metric ranges between 0 and 1 (higher values denote better performance; see Methods section for further details). Abbreviations: RF, random forest; AUC, area under the receiver operating characteristic curve; SCI, subjective cognitive impairment; CU, cognitively unimpaired; MCI, mild cognitive impairment; AD, Alzheimer’s disease.

ISSN: 1471-2318