Sir Ender, Aydogan Sena, Batur Sir Gul Didem, Celenlioglu Alp Eren
Department of Algology and Pain Medicine, University of Health Sciences Gulhane School of Medicine, Ankara, Turkey.
Department of Industrial Engineering, Gazi University, Ankara, Turkey.
J Pain Res. 2025 Jun 7;18:2839-2848. doi: 10.2147/JPR.S521331. eCollection 2025.
This study aims to use machine learning (ML) to explore predictive parameters related to the efficacy of caudal epidural pulsed radiofrequency (CEPRF) treatment for coccygodynia.
Five different ML methods were used to predict treatment success at 6 months after CEPRF. The findings generated by these algorithms are compared with respect to the accuracy of the results.
Symptom duration, angular deformation and NRS at admission are the most significant factors impacting therapy success in coccygodynia patients. Success rates are obtained for relatively short symptom durations to be 71.83%, for longer periods to be 16.67%; for short durations together with no angular deformity to be 79.55%, with angular deformity to be 59.26%; and for NRS level at admission less than 8 together with angular deformity to be 91.67%, with no angular deformity to be 33.33%.
This research reveals the potential of ML methods to improve treatment outcome prediction in coccygodynia. When a new patient is admitted, the ML-generated decision trees provide a quick and precise assessment of the possible success rate of CEPRF treatment.
本研究旨在运用机器学习（ML）探索与尾骶部硬膜外脉冲射频（CEPRF）治疗尾骨痛疗效相关的预测参数。
采用五种不同的机器学习方法预测CEPRF治疗6个月后的治疗成功率。将这些算法得出的结果在结果准确性方面进行比较。
症状持续时间、角变形和入院时的数字评分量表（NRS）是影响尾骨痛患者治疗成功的最重要因素。症状持续时间较短时的成功率为71.83%，较长时为16.67%；症状持续时间短且无角变形时为79.55%，有角变形时为59.26%；入院时NRS水平小于8且有角变形时为91.67%，无角变形时为33.33%。
本研究揭示了机器学习方法在改善尾骨痛治疗结果预测方面的潜力。当有新患者入院时，机器学习生成的决策树可对CEPRF治疗的可能成功率提供快速而精确的评估。
