Challenge Results

What a fantastic AI4Pain Grand Challenge 2025! This year we brought the community together across 32 teams, 20 submissions, 12 accepted papers, and 17 countries, all working to push the boundaries of physiological-signal–based pain assessment. Held in conjunction with the 27th International Conference on Multimodal Interaction (ICMI 2025), the challenge showcased cutting-edge models with real potential to transform clinical pain assessment. 

Thank you to everyone who trained, tuned, and shared insights, your work made this edition a standout. Every participant is a winner in our eyes, your effort, openness, and rigour moved the field forward. 

Congratulations to the top-performing teams on the hidden test set:

1. Team UniZar (University of Zaragoza), achieving an accuracy of 70.11%. Link to their paper: [PDF]

2. Team Takaathur (University of New South Wales & Nagoya University), reaching an accuracy of 62.93%. Link to their paper: [PDF]

Honourable Mention - Most Innovative Paper:

Team HMU_EDA (Hellenic Mediterranean University) with their paper "Multi-Representation Diagrams for Pain Recognition: Integrating Various Electrodermal Activity Signals into a Single Image". Link to their paper: [PDF]

The complete performance comparison of all participating teams on the hidden test set is presented in the figure below. The plot highlights the differences in model accuracy and demonstrates the overall progress achieved in physiological-based pain assessment this year.  

For more details on the challenge and the methodologies, please refer to the paper: 

Fernandez-Rojas, R., Joseph, C., Hirachan, N., Seymour, B. and Goecke, R., 2025, October. The ai4pain grand challenge 2025: Advancing pain assessment with multimodal physiological signals. In Companion Proceedings of the 27th International Conference on Multimodal Interaction (pp. 147-152).  [PDF]

We extend our deepest appreciation to all participating researchers for their remarkable contributions. Your innovations are accelerating the future of AI-driven pain assessment and demonstrating the promise of multimodal physiological sensing to improve healthcare outcomes. 

Stay tuned — the next edition of the AI4Pain Grand Challenge will be announced soon, and we look forward to welcoming you to AI4Pain 2026!