EEG and ECG Wearable Biosensor-Based Affective State Analysis Using Deep Learning over 6G IoT Healthcare Networks

Figures

• 

  Figure 1. Two-dimensional emotional model.

• 

  Figure 2. Illustration of a typical ECG beat morphology (P–QRS–T).

• 

  Figure 3. High-level deep learning classification pipelines for EEG- and ECG-based affective state recognition.

• 

  Figure 4. System architecture of the proposed framework.

• 

  Figure 5. Flow diagram of the modified CNN + attention mechanism for SEED emotional state classification.

• 

  Figure 6. Flow diagram of the modified ResNet50 architecture for WESAD stress classification.

• 

  Figure 7. 3D geometric model of a virtual hospital room.

• 

  Figure 8. Methodological framework for the deterministic channel modeling of a 6G-based virtual hospital room.

• 

  Figure 9. The results of emotional state classification using EEG signals. The label −1 represents negative emotions, 0 represents neutral emotions, and 1 represents positive emotions. The blue-colored diagonal cells represent the numbers of correctly classified samples, whereas the off-diagonal cells indicate the numbers of misclassified samples.

• 

  Figure 10. Confusion matrix for the ResNet50-based stress classifier on the WESAD dataset. The matrix illustrates the distribution of true versus predicted class labels across the three affective states.

• 

  Figure 11. Close-in reference distance models for the 7 GHz band (7.75 GHz carrier frequency) and the 92 GHz band (93 GHz carrier frequency) for the virtual hospital room.

• 

  Figure 12. 2D spatial distribution of received signal strength indicator (RSSI) for the virtual hospital room: (a) 7 GHz band (7.75 GHz carrier frequency); (b) 92 GHz band (93 GHz carrier frequency). The yellow star indicates the hotspot location, while the red circle denotes the boundary corresponding to the receiver sensitivity threshold of −80 dBm.