Applying machine learning to high-dimensional proteomics datasets for the identification of Alzheimer’s disease biomarkers

Table 9 Performance of machine learning models on protein ventricular data for Tasks A1, B1, and B2

	Accuracy	\(F_1\)-score	AUC	MCC
Task A1
XGB	0.76 (± 0.015)	0.31 (± 0.11)	0.70 (± 0.05)	0.17 (± 0.10)
LR	0.75 (± 0.03)	0.33 (± 0.15)	0.73 (± 0.06)	0.18 (± 0.16)
RF	0.75 (± 0.03)	0.19 (± 0.07)	0.71 (± 0.05)	0.07 (± 0.09)
Soft Ensemble	0.76 (± 0.02)	0.28 (± 0.09)	0.73 (± 0.04)	0.14 (± 0.09)
Hard Ensemble	0.77 (± 0.03)	0.28 (± 0.10)	0.73 (± 0.04)	0.15 (± 0.10)
Task B1
XGB	0.77 (\(\pm 0.04\))	0.35 (\(\pm 0.11\))	0.74 (\(\pm 0.07\))	0.21 (\(\pm 0.13\))
LR	0.77 (± 0.03)	0.37 (± 0.12)	0.66 (± 0.07)	0.23 (± 0.13)
RF	0.79 (± 0.03)	0.39 (± 0.11)	0.74 (± 0.07)	0.27 (± 0.12)
Soft ensemble	0.78 (± 0.04)	0.37 (± 0.13)	0.73 (± 0.07)	0.23 (± 0.14)
Hard ensemble	0.78 (± 0.04)	0.35 (± 0.13)	0.73 (± 0.07)	0.23 (± 0.15)
Task B2
XGB	0.78 (± 0.03)	0.53 (± 0.07)	0.81 (± 0.05)	0.39 (± 0.09)
LR	0.77 (± 0.03)	0.49 (± 0.07)	0.80 (± 0.05)	0.34 (± 0.07)
RF	0.79 (± 0.03)	0.51 (± 0.07)	0.81 (± 0.04)	0.38 (± 0.09)
Soft ensemble	0.79 (± 0.04)	0.54 (± 0.07)	0.82 (± 0.04)	0.41 (± 0.10)
Hard ensemble	0.79 (± 0.03)	0.54 (± 0.07)	0.82 (± 0.04)	0.41 (± 0.09)

This table shows the predictive performance of the best models on protein ventricular data across tasks A1, B1, and B2. Metrics reported include accuracy, \(F_1\)-score, AUC, and MCC, each with a 95% confidence interval. Results compare individual models (XGBoost, Logistic Regression, Random Forest) and ensemble methods (Soft and Hard Ensembles), highlighting variations in predictive accuracy and robustness across tasks

ISSN: 2045-8118