Table 3 Comparison with the state of the art pain classification approaches.
From: Decoding of pain during heel lancing in human neonates with EEG signal and machine learning approach
Source | Classified states | Method | Accuracy |
|---|---|---|---|
Vaart et al.24 | Noxious VS. non-noxious | Heart rate, Oxygen saturation, EEG, Facial, HR, saturation | 81% |
ML van der Vaart4 | Noxious VS. non-noxious | Behavioural and physiological (RR trace, EEG, etc.) | 76% 84% |
Nezam et al.12 | 3 Levels | Fractal dimension Shannon entropy approximate entropy spectral entropy | 83% |
Proposed method | No-pain VS. Pain | Kurtosis Shannon entropy power variance skewness | 100% |
Proposed method | 3 Levels | Kurtosis Shannon entropy power variance skewness | 84.2% |
