Table 1 Architectural Configuration of the Proposed Framework.

From: A hybrid reinforcement learning and knowledge graph framework for financial risk optimization in healthcare systems

Module

Layer Type

Input Dim

Output Dim

# Parameters

Patient Encoder (VAE)

Dense Layer 1

18

64

1,216

Dense Layer 2 (ReLU)

64

32

2,080

Latent Layer \(\mu\), \(\sigma\)

32

16 each

1,056

Categorical Embeddings

Region Embedding

4

4

16

Insurance Embedding

5

5

25

Medical Condition Embedding

10

6

60

R-GCN (Knowledge Graph)

GCN Layer 1

128

64

8,192

GCN Layer 2 (ReLU)

64

32

2,048

Policy Network (PPO / DQN)

Input Layer

90

128

11,648

Hidden Layer 1 (ReLU)

128

64

8,256

Hidden Layer 2 (Dropout 0.3)

64

32

2,080

Output Layer (Softmax/Q)

32

5

165

Total Parameters

   

36,842