Background & Aims

Conditioned pain modulation (CPM) serves as an experimental psychophysical tool to assess the human endogenous pain inhibition system. In healthcare, machine learning methods are increasingly utilized to enhance clinical decision-making. These algorithms autonomously learn from data, identifying key variables that can predict outcomes without explicit programming. Our aims were twofold: (i) to train and test various machine learning models, focusing on their ability to estimate the functionality of endogenous pain inhibitory pathways in individuals suffering from musculoskeletal pain, and (ii) explore the external validation of the model with a new dataset.

Methods

Patients with musculoskeletal pain were recruited from outpatient services. Data collection involved sociodemographic, lifestyle, and clinical aspects. CPM efficacy was measured by comparing pressure pain thresholds pre- and post-immersion of the non-dominant hand in 1–4 °C water (cold-pressure test). We developed five ML models: decision tree, random forest, gradient-boosted trees, logistic regression, and support vector machine. Model performance was evaluated using AUC, accuracy, sensitivity, specificity, precision, recall, F1-score, and MCC. SHapley Additive explanation values and LIME were used for model interpretation.

Results

The first objective of our study was completed, revealing that the XGBoost model showcased the highest performance, marked by an accuracy of 0.81 (95% CI: 0.73 to 0.89), F1 score of 0.80 (95% CI: 0.74 to 0.87), AUC of 0.81 (95% CI: 0.74 to 0.88), along with MCC and Kappa both at 0.61. Influential factors for the model included pain duration, fatigue, physical activity, and the count of painful areas. The external validation of the model (second aim) is currently under progress (data analysis).

Conclusions

XGBoost showed potential in predicting the CPM efficacy in patients with musculoskeletal pain on our dataset. Further research is needed to ensure the external validity and clinical utility of these models.

References

1. Arendt-Nielsen L, Skou ST, Nielsen TA, Petersen KK. Altered Central Sensitization and Pain Modulation in the CNS in Chronic Joint Pain. Current Osteoporosis Reports. 2015;13(4):225–34.

2. Arendt-nielsen L, Morlion B, Perrot S, Dahan A, Dickenson A, Kress HG, et al. Assessment and manifestation of central sensitisation across different chronic pain conditions. European Journal of Pain. 2018;22(2):216–41.

3. Ossipov MH, Morimura K, Porreca F. Descending pain modulation and chronification of pain. Current opinion in supportive and palliative care. 2014;8(2):143.

4. Yarnitsky D, Bouhassira D, Drewes AM, Fillingim RB, Granot M, Hansson P, et al. Recommendations on practice of conditioned pain modulation (CPM) testing. European Journal of Pain (United Kingdom). 2015;19(6):805–6.

5. Ramaswamy S, Wodehouse T. Conditioned pain modulation —– A comprehensive review. Neurophysiologie Clinique/Clinical Neurophysiology. 2021;51(3):197–208.

6. Lewis GN, Heales L, Rice DA, Rome K, McNair PJ. Reliability of the conditioned pain modulation paradigm to assess endogenous inhibitory pain pathways. Pain Research and Management. 2012;17(2):98–102.

7. Clark J, Nijs J, Yeowell G, Goodwin PC. What are the predictors of altered central pain modulation in chronic musculoskeletal pain populations? A Systematic review. Pain physician. 2017;20(6):487–500.

8. Lewis GN, Rice DA, McNair PJ. Conditioned pain modulation in populations with chronic pain: a systematic review and meta-analysis. The journal of pain. 2012;13(10):936–44.

9. Lötsch J, Ultsch A. Machine learning in pain research. Pain. 2018;159(4):623.

10. Hastie T, Tibshirani R, Friedman J. The elements of statistical learning: data mining, inference, and prediction. Springer Science & Business Media; 2009.

Presenting Author

Felipe J. J. Reis

Poster Authors

Felipe Reis

PhD

Federal Institute of Rio De Janeiro

Lead Author

Juliana Valentim Bittencourt

UNISUAM

Lead Author

Arthur de Sá Ferreira

UNISUAM

Lead Author

Ney Meziat-Filho

UNISUAM

Lead Author

Leandro Calazans Nogueira

UNISUAM

Lead Author

Topics

  • Assessment and Diagnosis