Table 8 Ablation experiments results of three different feature extraction approaches. The best performance is denoted as bold.

From: Decoding potential lncRNA and disease associations through graph representation learning and gradient boosting with histogram

Dataset

Feature

Precision

Recall

Accuracy

F1-score

AUC

AUPR

lncRNADisease v2.0

Linear feature

0.8746 ± 0.0327

0.7997 ± 0.0399

0.8420 ± 0.0259

0.8347 ± 0.0283

0.9200 ± 0.0192

0.9292 ± 0.0177

Nonlinear feature

0.8457 ± 0.0252

0.8581 ± 0.0317

0.8503 ± 0.0196

0.8513 ± 0.0199

0.9244 ± 0.0140

0.9193 ± 0.0188

LDA-GMCB

0.8842 ± 0.0235

0.8707 ± 0.0289

0.8781 ± 0.0199

0.8771 ± 0.0206

0.9464 ± 0.0135

0.9506 ± 0.0143

MNDR

Linear feature

0.9174 ± 0.0137

0.9038 ± 0.0168

0.9111 ± 0.0103

0.9104 ± 0.0105

0.9688 ± 0.0057

0.9733 ± 0.0045

Nonlinear feature

0.9138 ± 0.0209

0.9095 ± 0.0202

0.9117 ± 0.0167

0.9115 ± 0.0166

0.9708 ± 0.0243

0.9703 ± 0.0319

LDA-GMCB

0.9339 ± 0.0131

0.9155 ± 0.0168

0.9253 ± 0.0105

0.9245 ± 0.0108

0.9734 ± 0.0058

0.9779 ± 0.0046

lncRNADisease v3.0

Linear feature

0.8961 ± 0.0133

0.8734 ± 0.0156

0.8859 ± 0.0095

0.8845 ± 0.0098

0.9536 ± 0.0061

0.9550 ± 0.0068

Nonlinear feature

0.8915 ± 0.0142

0.9073 ± 0.0161

0.8983 ± 0.0107

0.8992 ± 0.0107

0.9619 ± 0.0060

0.9583 ± 0.0072

LDA-GMCB

0.9060 ± 0.0135

0.9163 ± 0.0121

0.9105 ± 0.0094

0.9110 ± 0.0092

0.9657 ± 0.0060

0.9605 ± 0.0086

Back to article page