Çok Modlu Biyosinyallerle Akut Ağrıların Makine Öğrenmesiyle Tespiti

Yıl 2025, Cilt: 6 Sayı: 2, 85 - 100, 19.10.2025

Cüneyt Özdemir

https://doi.org/10.70562/tubid.1736747

Öz

Ağrının subjektif doğası, nesnel değerlendirme yöntemlerinin geliştirilmesini zorunlu kılmaktadır. Bu çalışma, giyilebilir sensörlerden elde edilen çok modlu fizyolojik sinyalleri kullanarak farklı akut ağrı tiplerini otomatik olarak sınıflandırmanın fizibilitesini ve etkinliğini araştırmaktadır. Kamuoyuna açık "PhysioPain Dataset"ten alınan ve 99 katılımcıya ait baş ağrısı, sırt ağrısı, menstrüel ağrı ve ağrısız du- rumlardaki fizyolojik veriler (BVP, EDA, ACC, TEMP) analiz edilmiştir. Ham zaman serisi verilerinden, istatistiksel ve frekans tabanlı ayırt edici özellikler çıkarılmış ve bu özellikler, Rastgele Orman, XGBoost ve Destek Vektör Makineleri gibi standart makine öğrenmesi modelleriyle sınıflandırılmıştır. Deneysel sonuçlar, topluluk tabanlı modellerin olağanüstü bir performans sergilediğini ortaya koymuştur. XGBo- ost ve Rastgele Orman modelleri, test seti üzerinde sırasıyla %99 ve %98’lik doğruluk ve F1-skorlarına ulaşarak, çıkarılan özelliklerin farklı ağrı durumlarını ayırt etmede son derece etkili olduğunu kanıtla- mıştır. Bu yüksek başarı, daha karmaşık derin öğrenme mimarilerine olan ihtiyacı ortadan kaldırmıştır. Sonuç olarak, bu çalışma, etkili bir özellik mühendisliği süreci ve standart makine öğrenmesi algoritmaları ile fizyolojik sinyallerden akut ağrı tiplerinin neredeyse mükemmel bir doğrulukla tespit edilebileceğini göstermektedir. Bu bulgular, giyilebilir teknolojilerin ağrının nesnel, otomatik ve non-invaziv takibi için güvenilir bir klinik araç olma potansiyelini güçlendirmektedir.

Anahtar Kelimeler

Ağrı Değerlendirmesi , Makine Öğrenmesi , Fizyolojik Sinyaller , Sinyal İşleme , Giyilebilir Sensörler

Detection of Acute Pain with Multimodal Biosignals Using Machine Learning

Öz

The subjective nature of pain necessitates the development of objective assessment methods. This study investigates the feasibility and efficacy of automatically classifying different types of acute pain using multimodal physiological signals obtained from wearable sensors. Physiological data (BVP, EDA, ACC, TEMP) from 99 participants, encompassing headache, back pain, menstrual pain, and pain-free states, were analyzed using the publicly available "PhysioPain Dataset." Discriminative statistical and frequency-based features were extracted from raw time-series data and classified using standard machine learning models, including Random Forest, XGBoost, and Support Vector Machines. Experimental results demonstrate that ensemble-based models exhibited exceptional performance. Specifically, XGBoost and Random Forest achieved accuracy and F1-scores of 99% and 98%, respectively, on the test set, confirming the high effectiveness of the extracted features in distinguishing various pain states. This superior performance eliminates the need for more complex deep learning architectures. In conclusion, this study demonstrates that acute pain types can be detected with near-perfect accuracy through effective feature engineering and standard machine learning algorithms applied to physiological signals. These findings reinforce the potential of wearable technologies as reliable clinical tools for the objective, automated, and non-invasive monitoring of pain

Anahtar Kelimeler

Pain Assessment , Machine Learning , Physiological Signals , Signal Processing , Wearable Sensors.

Kaynakça

• 1. Loeser JD, Treede R-D. The kyoto protocol of IASP basic pain terminology. Pain. 2008;137(3):473–7.
• 2. Breivik H, Collett B, Ventafridda V, Cohen R, Gallacher D. Survey of chronic pain in Europe: Prevalence, impact on daily life, and treatment. Eur J Pain. 2006;10(4):287–333.
• 3. Wong DL, Baker CM. Pain in children: Comparison of assessment scales. Pediatr Nurs. 1988;14(1):9–17.
• 4. Werner P, Al-Hamadi A, Niese R, Walter S, Gruss S, Traue HC. Towards pain monitoring: Facial expression, head pose, a new database, an initial evaluation and a roadmap. PLoS One. 2014;9(10):e110188.
• 5. Raja SN, Carr DB, Cohen M, Finnerup NB, Flor H, Gibson S, et al. The revised international association for the study of pain definition of pain: Concepts, challenges, and compromises. Pain. 2020;161(9):1976–82.
• 6. Brahnam S, Nanni L, McMurtrey S, Lumini A, Brattin R, Slack M, et al. Neonatal pain detection in videos using the ICOPEVID dataset and an ensemble of descriptors extracted from Gaussian of local descriptors. Appl Comput Inform. 2023;19(1–2):122–43.
• 7. Von Baeyer CL, Children’s self-reports of pain intensity: scale selection, limitations and interpretation. Pain Res Manag. 2006;11(3):157–62.
• 8. Ajayi TA, Salongo L, Zang Y, Wineinger N, Steinhubl S. Mobile health-collected biophysical markers in children with serious illness-related pain. J Palliat Med. 2021;24(4):580–8.
• 9. Walter S, Gruss S, Ehleiter H, Tan J, Traue HC, Werner P, et al. The BioVid heat pain database: Data for the advancement and systematic validation of an automated pain recognition system. In: 2013 IEEE International Conference on Cybernetics (CYBCO). IEEE; 2013. p. 128–31.
• 10. Lucey P, Cohn JF, Prkachin KM, Solomon PE, Matthews I. Painful data: The UNBC-McMaster shoulder pain expression archive database. In: 2011 IEEE Int Conf on Automatic Face & Gesture Recognition (FG). IEEE; 2011. p. 57–64.
• 11. Rodriguez P, Cucurull G, Gonzàlez J, Gonfaus JM, Nasrollahi K, Moeslund TB, et al. Deep pain: Exploiting long short-term memory networks for facial expression classification. IEEE Trans Cybern. 2017;52(5):3314–24.
• 12. Subramanian A, Cao R, Naeini EK, Aqajari SAH, Hughes TD, Calderon M-D, et al. Multimodal pain recognition in postoperative patients: Machine learning approach. JMIR Form Res. 2025;9:e67969.
• 13. Phan KN, Iyortsuun NK, Pant S, Yang HJ, Kim SH. Pain recognition with physiological signals using multi-level context information. IEEE Access. 2023;11:20114–27.
• 14. Cazzola M, Atzeni F, Boccassini L, Cassisi G, Sarzi-Puttini P, et al. Physiopathology of pain in rheumatology. Reumatismo. 2014;66(1):4–13.
• 15. Barut K, Adrovic A, Şahin S, Kasapçopur Ö. Juvenile idiopathic arthritis. Balkan Med J. 2017;34(2):90–101.
• 16. Empatica Inc. Empatica E4 wristband technical specifications [Internet]. 2019 [cited 2025 Jul 30]. Available from: https://www.empatica.com/research/e4/
• 17. Storm H. Changes in skin conductance as a tool to monitor nociceptive stimulation and pain. Curr Opin Anaesthesiol. 2002;15(5):609–14.
• 18. Loggia ML, Mogil JS, Bushnell MC. Empathy hurts: Compassion for another increases both sensory and affective components of pain perception. Pain. 2011;152(5):1049–56.
• 19. Paras M, Rudzinska AK. Motion analysis in pain research. Pain. 2009;141(1–2):1–2.
• 20. Herborn KA, Graves JL, Jerem P, Evans NP. Skin temperature reveals the intensity of acute stress. Physiol Behav. 2015;152:225–30.
• 21. Subramanian R, Wager TD, Mackey S, Reiss AL. Multimodal pain assessment using physiological signals. J Pain Res. 2018;11:2237–47.
• 22. Thiam P, Kessler V, Walter S, Traue HC. Multimodal pain recognition in the context of human-robot interaction. Front Neurosci. 2019;13:1034.
• 23. Werner P, Lopez-Martinez D, Walter S, Al-Hamadi A, Gruss S, Picard RW. Automatic recognition methods supporting pain assessment: A survey. IEEE Trans Affect Comput. 2019;13(1):430–48.
• 24. Orvile. Physiopain dataset: A multimodal dataset for pain assessment [Internet]. Kaggle; 2024 [cited 2025 Jul 30]. Available from: https://www.kaggle.com/datasets/orvile/physiopain-dataset
• 25. Rodriguez P, Cucurull G, Gonfaus JM, Roca FX, Gonzalez J. Deep pain: Exploiting long short-term memory networks for facial expression classification. IEEE Trans Cybern. 2022;52(5):3314–24.
• 26. Brahnam S, Nanni L, Sexton A, Lumini A. Neonatal pain detection in videos using the ICOPEVID dataset and an ensemble of descriptors extracted from gaussian of local descriptors. Appl Comput Inform. 2020 [Preprint].
• 27. Pinzon-Arenas JO, Kong Y, Chon KH, Posada-Quintero HF. Design and evaluation of deep learning models for continuous acute pain detection based on phasic electrodermal activity. IEEE J Biomed Health Inform. 2023;27(9):4250–60.
• 28. Subramanian A, Ganapathy N, Lázaro J, Perez-Ajates JA, Gil E, Joshi R. Multimodal pain recognition in postoperative patients: A machine learning approach. bioRxiv [Preprint]. 2023.
• 29. Yıldırım E. Objektif ağrı değerlendirmesi için çok boyutlu biyosinyal füzyonu. 2024.
• 30. Melzack R. The McGill pain questionnaire: Major properties and scoring methods. Pain. 1975;1(3):277–99.
• 31. Siddiqui MIH, Sakib AH, Akter S, Debnath J, Mahmud MR. Comparative analysis of traditional machine learning vs deep learning for sleep stage classification. Int J Sci Res Arch. 2025;1778–89.
• 32. Afuan L, Isnanto RR. A comparative study of machine learning algorithms for fall detection in technology-based healthcare system: Analyzing SVM, KNN, decision tree, random forest, LSTM, and CNN. E3S Web Conf. 2025;605:03051.
• 33. Chandragandhi S, Arvind C, Srihari K. Advanced predictive disease modeling in biomedical IoT using the temporal adaptive neural evolutionary algorithm. Sci Rep. 2025;15:20378.
• 34. Kılıç Ş. HybridVisionNet: An advanced hybrid deep learning framework for automated multi-class ocular disease diagnosis using fundus imaging. Ain Shams Eng J. 2025;16(10):103594.
• 35. Kılıç Ş. Attention-based dual-path deep learning for blood cell image classification using ConvNeXt and Swin Transformer. J Imaging Inform Med. 2025;1–19.
• 36. Çelik Y. Bellek tabanlı LSTM ve GRU makine öğrenmesi algoritmaları kullanarak BIST100 endeks tahmini. Fırat Üniv Müh Bilim Derg. 2024;36(2):553–61.
• 37. Celik Y, Karabatak M. Extracting low dimensional representations from large size whole slide images using deep convolutional autoencoders. Expert Syst. 2023;40(4):e12819.
• 38. Kılıç Ş, Doğan, Y. Deep learning based gender identification using ear images. Traitement du Signal, 20223;40(4).