Table 2 Performance of various AI models for colorectal cancer (CRC) risk prediction over 1–5 year horizons

From: Deep multimodal fusion of patho-radiomic and clinical data for enhanced survival prediction for colorectal cancer patients

Architectural Class

Model

Risk Prediction (AUC)

C-index

WMAE

  

1y

2y

3y

4y

5y

  

Graph Neural Network

DM-GNN41

0.70

0.71

0.69

0.68

0.65

0.67

2.94

Graph Neural Network

SAGL42

0.75

0.73

0.70

0.73

0.64

0.69

2.80

Spatiotemporal GNN

STG43

0.78

0.65

0.67

0.69

0.71

0.70

2.84

Deep Neural Network

DeepCRC44

0.77

0.75

0.70

0.62

0.69

0.71

2.37

Multimodal Hypergraph

MRePath45

0.78

0.79

0.76

0.73

0.74

0.72

1.64

AI-Augmented DL

Risk-Net46

0.81

0.78

0.74

0.75

0.76

0.74

1.83

Ours

0.85

0.89

0.84

0.79

0.80

0.82

1.40

  1. More advanced models show higher time-dependent AUC-ROC and C-index, and lower WMAE. WMAE evaluates survival prediction error, accounting for censoring.
Back to article page