AI4Pain Assessment

This repository contains the implementation of our AI4Pain pain-level assessment framework using physiological signals.

This work is associated with our paper accepted/published in AICOMS 2026, currently in press with Springer.

Project Overview

The goal of this project is to classify pain levels using multimodal physiological signals such as:

• EDA
• BVP
• RESP
• SpO₂

The framework includes preprocessing, modality-specific modeling, attention-based fusion, and final pain-level classification.

Repository Structure

preprocessing/   Data preprocessing scripts
src/models/          Model architecture files
src/training/        Training scripts
src/evaluation/      Evaluation and testing scripts
Figures/             Prediction files and figures
checkpoints/         Saved model weights
data/                Raw/processed/sample data

Dataset

We used tha datset from this challenge: https://sites.google.com/view/ai4pain2025/challenge-details?authuser=0 Inside the data folder, We have added some example of data's. You can the idea of data from there.

Citation

If you use this code, please cite our AICOMS 2026 paper:

@inproceedings{bappi2026ai4pain,
  title={AI4Pain Assessment Using Multimodal Physiological Signals},
  author={Bappi, Ilias and others},
  booktitle={AICOMS 2026},
  publisher={Springer},
  year={2026},
  note={In press}
}