Table 3 Comparative analysis of the current study results with state-of-the-art pain level classifications.
Study | Year | Data | Pain source | Subjects | Samples | Number of classes | Model | Test samples | Accuracy (%) |
|---|---|---|---|---|---|---|---|---|---|
Rojas et al.83 | 2021 | fNIRS | Thermal pain perceptions | 18 healthy subjects | 300 | 2 (cold and hot induced pain) | Deep learning (bidirectional long short-term memory) | 30% of randomly selected samples | 90.6 |
Fernandez Rojas et al.47 | 2017 | fNIRS | Thermal pain perceptions | 18 healthy subjects | 352 | 2 (cold and hot induced pain) | k-nearest neighborhood | Data from 5 subjects | 92.08 |
Lopez-Martinez et al.14 | 2019 | fNIRS | Electrical noxious stimuli and tactile brush stimuli | 43 healthy participants | 43 | 2 (no pain and pain) | Support Vector Machine (rbf kernel) | 10% of randomly selected sampled | 69 |
Current study | 2025 | fNIRS | Cancer-related pain | 93 cancer patients and 13 healthy participants | 129 | 3 (no or mild pain, moderate pain, severe pain) | Multinomial logistic regression | 129 test samples obtained by applying leave-one-participant-out cross validation | 74 |
