AMELİYAT SONRASI KRONİK AĞRIDA AKILLI BİR ÖNGÖRÜ SİSTEMİ (I-POCPP)

Abstract

Amaç: Ameliyat Sonrası Kronik Ağrı (Postoperative Chronic Pain - POCP), hastaların yaşam kalitesini etkilemektedir. Makine öğrenmesi ve uygulamaları, ağrı araştırmalarına önemli katkılar sağlamaktadır. En iyi performans gösteren makine öğrenmesi algoritmasını kullanarak “Ameliyat Sonrası Kronik Ağrıda Akıllı Bir Öngörü Sistemi (I-POCPP)” geliştirerek perioperatif verilere dayalı olarak hastaların ameliyat sonrası kronik ağrı durumunu öngörmek hedeflenmiştir.
Gereç ve Yöntem: Bu çok merkezli çalışmanın veri seti, Türkiye’deki üçüncü basamak 5 hastanede elektif koşullarda anestezi altında ameliyat olan 733 hastadan toplanmıştır. Çalışmada farklı makine öğrenmesi öngörü algoritmaları kullanılmıştır. Anestezistler tarafından tanı konulan hastaların gerçekleşen kronik ağrı durumu ve modellerin öngörü sonuçları karşılaştırılarak modellerin performansı değerlendirilmiştir.
Bulgular: k-En Yakın Komşu (kNN), Rastgele Orman (RF) ve C5.0 modellerinin bir hastanın ameliyat sonrası kronik ağrı durumunu %80’den yüksek doğrulukla öngörebildiği bulunmuştur. kNN algoritmasının kararlı bir modeli olmadığı düşüncesiyle RF performansı dikkate alınmıştır. RF ve Sınıflandırma ve Regresyon Ağacı (CART) algoritmalarının nitelik önem sıralamasına göre “Kesi yeri”, “Yaş” ve “Ameliyat nedeni” ortaktır. C5.0 algoritması sonucunda elde edilen nitelik önem sıralaması RF ve CART modelleri ile uyumlu olmadığı için bu modelin sonuçları değerlendirilmemiştir. Tüm modeller arasında en iyi sonuç RF ile elde edilmiştir ve buna göre I-POCPP geliştirilmiştir. Sonuç: I-POCPP sistemiyle sağlanan ameliyat sonrası kronik ağrının hızlı, doğru ve etkin tedavisi, hastanın günlük yaşama daha erken dönmesini sağlayabilir.

Keywords

• Ameliyat Sonrası Kronik Ağrı
• Makine Öğrenmesi
• Sınıflandırma
• Karar Destek Sistemi.

AN INTELLIGENT POSTOPERATIVE CHRONIC PAIN PREDICTION SYSTEM (I-POCPP)

Abstract

Objective: Postoperative Chronic Pain (POCP) affects the quality of patients’ lives. Machine learning and its applications provide significant contributions to pain research. The aim of this study is to predict the POCP status of patients based on perioperative data by developing an “Intelligent POCP Prediction System (I-POCPP)” using the best performing machine learning algorithm.
Material and Method: The dataset for this multi-centered study was collected from 5 tertiary hospitals in Turkey and included 733 patients who had undergone elective surgeries attended by an anesthesiologist in the operating room. Several machine learning prediction algorithms were used. POCP status of the patients diagnosed by the anesthesiologists and the prediction results of the models were compared to evaluate the performance of the models.
Results: It was found that the k-Nearest Neighbour (kNN), Random Forest (RF), and C5.0 models were able to predict the POCP status of a patient with an accuracy higher than 80%. The performance of RF was considered, while the kNN algorithm has no stable model. According to RF and Classification and Regression Tree (CART) algorithms’ attribute importance ranking, “Incision site”, “Age”, and “Primary diagnosis for operation” are common attributes. Since the attribute importance ranking obtained as a result of the C5.0 algorithm was not consistent with the RF and CART models, the results of this model were not evaluated. The best result among all models was obtained by RF, and I-POCPP has been developed accordingly. Conclusion: Fast, accurate, and efficient treatment of POCP provided by I-POCPP could allow the patient to return to daily life earlier.

Keywords

• Post-Operative Chronic Pain
• Machine Learning
• Classification
• Decision Support System

References

1. 1. Macrae WA. Chronic post-surgical pain: 10 years on. Br J Anaesth 2008;101(1):77-86. [CrossRef] google scholar
2. 2. Correll D. Chronic postoperative pain: recent findings in understanding and management. F1000Res 2017;6:1054. [CrossRef] google scholar
3. 3. Lötsch J, Ultsch A. Machine learning in pain research. Pain. 2018;159(4):623-30. [CrossRef] google scholar
4. 4. Shearer C. The CRISP-DM model: the new blueprint for data mining. J Data Warehous 2000;5(4):13-22. google scholar
5. 5. Balaban ME, Kartal E. Veri Madenciliği ve Makine Öğrenmesi Temel Algoritmaları ve R Dili ile Uygulamaları. 2nd ed. Beyoğlu, İstanbul: Çağlayan Kitabevi; 2018. google scholar
6. 6. Sutaş Bozkurt AP, Özen Z, Kartal E, Emre İE, Selçukcan Erol Ç, Koçoğlu FÖ, et al. Analysis of the incidence and predictive factors of chronic postoperative pain in adult population. Tepecik Eğitim ve Araşt Hastan Derg 2018;28(2):89-94. [CrossRef] google scholar
7. 7. Babar V, Ade R. MLP-based undersampling technique for imbalanced learning. In: Proceedings of International Conference on Automatic Control and Dynamic Optimization Techniques (ICACDOT); 2016 Sept 9-10. Pune, India: IEEE; 2016. p. 142-7. [CrossRef] google scholar
8. 8. The R Foundation. R: The R Project for Statistical Computing. 2019 (cited 2019 Dec 19). Available from: URL: https://www.r-project.org/ google scholar

9. 9. RStudio. RStudio - Open source and enterprise-ready professional software for R. 2018 (cited 2018 Feb 10). Available from: URL: https://www.rstudio.com/ google scholar
10. 10. RStudio. Shiny. 2017 (cited 2019 Feb 4). Available from: URL: https://shiny.rstudio.com/ google scholar
11. 11. RStudio. shinyapps.io. 2017 (cited 2019 Feb 4). Available from: URL: http://www.shinyapps.io/ google scholar
12. 12. Louridas P, Ebert C. Machine Learning. IEEE Softw 2016;33(5):110-5. [CrossRef] google scholar
13. 13. Hsu YW, Somma J, Hung YC, Tsai PS, Yang CH, Chen CC. Predicting postoperative pain by preoperative pressure pain assessment. Anesthesiology 2005;103(3):613-8. [CrossRef] google scholar
14. 14. Rotboll Nielsen P, Rudin Â, Werner MU. Prediction of postoperative pain. Curr Anaesth Crit Care 2007;18(3):157-65. [CrossRef] google scholar
15. 15. Werner MU, Mjöbo HN, Nieİsen PR, Rudin A. Prediction of postoperative pain: a systematic review of predictive experimentaİ pain studies. Anesthesioİogy 2010;112(6):1494-502. [CrossRef] google scholar
16. 16. Persson AKM. Predicting postoperative pain. Cİinicaİ and genetic studies of reİationships between pain sensitivity and pain after surgery. Denmark: Lund Univ. DoctoraİThesis. 2018. google scholar
17. 17. Tighe PJ, Lucas SD, Edwards DA, Boezaart AP, Aytug H, Bihorac A. Use of machine-İearning cİassifiers to predict requests for preoperative acute pain service consuİtation. Pain Med 2012 Oct;13(10):1347-57. [CrossRef] google scholar
18. 18. Nickerson P, Tighe P, Shickeİ B, Rashidi P. Deep neuraİ network architectures for forecasting anaİgesic response. Annu Int Conf IEEE Eng Med Bioİ Soc 2016;2016:2966-9. [CrossRef] google scholar
19. 19. Garcia-Chimeno Y, Garcia-Zapirain B, Gomez-Beİdarrain M, Fernandez-Ruanova B, Garcia-Monco JC. Automatic migraine cİassification via feature seİection committee and machine İearning techniques over imaging and questionnaire data. BMC Med Inform Decis Mak 2017;17(1):38. [CrossRef] google scholar
20. 20. Lötsch J, Sipila R, Dimova V, Kalso E. Machine-learned seİection of psychoİogicaİ questionnaire items reİevant to the development of persistent pain after breast cancer surgery. Br J Anaesth 2018;121(5):1123-32. [CrossRef] google scholar
21. 21. Banu A B, Thirumalaikolundusubramanian P. Comparison of Bayes Classifiers for Breast Cancer Classification. Asian Pac J Cancer Prev 2018;19(10):2917-20. google scholar
22. 22. Yousef M, Nebozhyn M, Shatkay H, Kanterakis S, Showe LC, Showe MK. Combining multi-species genomic data for microRNA identification using a Naive Bayes classifier. Bioinformatics 2006;22(11):1325-34. [CrossRef] google scholar
23. 23. Mohktar MS, Redmond SJ, Antoniades NC, Rochford PD, Pretto JJ, Basilakis J, Lovell NH, McDonald CF. Predicting the risk of exacerbation in patients with chronic obstructive pulmonary disease using home telehealth measurement data. Artif Intell Med 2015;63(1):51-9. [CrossRef] google scholar
24. 24. Venkatesan E, Velmurugan T. Performance Analysis of Decision Tree Algorithms for Breast Cancer Classification. Indian J Sci Technol 2015;8(29):1-8. [CrossRef] google scholar
25. 25. Salgueiro M, Basogain X, Collado A, Torres X, Bilbao J, Donate F, Aguilera L, Azkue JJ. An artificial neural network approach for predicting functional outcome in fibromyalgia syndrome after multidisciplinary pain program. Pain Med 2013;14(10):1450-60. [CrossRef] google scholar
26. 26. Szaleniec J, Szaleniec M, Strşk P. A stepwise protocol for neural network modeling of persistent postoperative facial pain in chronic rhinosinusitis. Bio-Algorithms Med-Syst 2016;12(2):81-8. [CrossRef] google scholar
27. 27. Tighe PJ, Harle CA, Hurley RW, Aytug H, Boezaart AP, Fillingim RB. Teaching a Machine to Feel Postoperative Pain: Combining High-Dimensional Clinical Data with Machine Learning Algorithms to Forecast Acute Postoperative Pain. Pain Med 2015;16(7):1386-401. [CrossRef] google scholar
28. 28. Tighe P, Laduzenski S, Edwards D, Ellis N, Boezaart AP, Aygtug H. Use of machine learning theory to predict the need for femoral nerve block following ACL repair. Pain Med 2011;12(10):1566-75. [CrossRef] google scholar