Skip to main content

Table 4 Performance investigation of the model with different numbers of GAT layers

From: MODILM: towards better complex diseases classification using a novel multi-omics data integration learning model

Dataset

Method

ACC

F1

AUC

F1_weighted

F1_macro

ROSMAP

2-layer MODILM (Ours)

0.843

0.850

0.891

–-

–-

3-layer MODILM

0.803

0.811

0.875

–-

–-

4-layer MODILM

0.731

0.756

0.839

–-

–-

LGG-2

2-layer MODILM (Ours)

0.975

0.978

0.993

–-

–-

3-layer MODILM

0.928

0.946

0.917

–-

–-

4-layer MODILM

0.849

0.883

0.919

–-

–-

BRCA

2-layer MODILM (Ours)

0.845

–-

–-

0.840

0.804

3-layer MODILM

0.809

–-

–-

0.795

0.713

4-layer MODILM

0.765

–-

–-

0.731

0.612

SKCM

2-layer MODILM (Ours)

0.928

–-

–-

0.927

0.925

3-layer MODILM

0.902

–-

–-

0.902

0.903

4-layer MODILM

0.894

–-

–-

0.895

0.894

LGG-4

2-layer MODILM (Ours)

0.954

–-

–-

0.954

0.948

3-layer MODILM

0.881

–-

–-

0.865

0.838

4-layer MODILM

0.795

–-

–-

0.769

0.748

LUSC

2-layer MODILM (Ours)

0.865

–-

–-

0.855

0.833

3-layer MODILM

0.768

–-

–-

0.687

0.626

4-layer MODILM

0.750

–-

–-

0.677

0.621