Reza SAFIARIAN, Payam AMİNİ, Khodayari Elham MOEZ, Fatemeh MOHAMMADZADEH, Farid ZAYERI, Mohammad TAVAKOLI

Risk group classification for bleeding after coronary artery bypass graft surgery: A comparison of the logistic regression with decision tree models

Amaç: Bu çalışmada koroner arter baypas greft (KABG) cerrahisi sonrasında kanama açısından yüksek riskli hasta grupları belirlendi. Çalışma planı: Haziran 2001 - Ağustos 2008 tarihleri arasında İran Tahran Jamaran Kalp Hastanesinde KABG cerrahisi yapılan 205 hasta (143 erkek, 62 kadın; ort. yaş 59.7±10.1 yıl; dağılım 28-83 yıl) retrospektif olarak değerlendirildi. Hastaların başlangıç- taki özellikleri ve ameliyat sonrası kanama durumları kaydedildi. Kanama olan ve kanama olmayan hastaları sınıflandırırken, klasik lojistik regresyon ve karar ağacı modelleri kullanıldı. Bulgular: Lojistik regresyon analizinde cinsiyetin ameliyat sonrası kanama ile anlamlı düzeyde ilişkili olduğu görüldü. Karar ağacı modelinde ise, yaş (skor= 100), diabetes mellitus (skor= 16.38), cinsiyet (skor= 13.67), başkentte ikamet etme (skor= 7.31) ve disli- pideminin (skor= 5.06) kanama üzerinde etkisi olduğu belirlendi. Lojistik regresyona kıyasla, karar ağacı modelinde hastaların daha iyi sınıflandırıldığı da gözlendi. Sonuç: Cerrahlar KABG öncesinde üç damarına baypas yapılan hastalarda ileri yaş, erkek cinsiyeti, diabetes mellitus yokluğu ve dislipidemi varlığı gibi kanamanın risk göstergelerini göz önünde bulundurmalıdır. Ayrıca, istatistik uzmanlarına risk grubu sınıflan- dırmasında lojistik regresyon analizinin yerine karar ağacı modelini kullanmalarını önermekteyiz.

Koroner arter baypas greft cerrahisi sonrasında kanama açısından riskli grupların sınıflandırılması: Lojistik regresyon ve karar ağacı modellerinin karşılaştırılması

Background: This study aims to identify high-risk patient groups for bleeding after coronary artery bypass graft (CABG) surgery. Methods: We retrospectively evaluated 205 patients (143 males, 62 females; mean age 59.7±10.1 years; range, 28 to 83 years) undergoing CABG surgery between June 2001 and August 2008 at Jamaran Heart Hospital, Tehran, Iran. Baseline characteristics of the patients and postoperative bleeding status were recorded. For classifying the bleeders and non-bleeders, classic logistic regression and decision tree models were utilized. Results: Logistic regression analysis showed that sex was significantly related to postoperative bleeding. Decision tree model revealed that age (score= 100), diabetes mellitus (score= 16.38), sex (score= 13.67), capital residency (score= 7.31) and dyslipidemia (score= 5.06) were found to have an impact on bleeding. We also observed that the decision tree model provided a better classification of the patients than logistic regression. Conclusion: Surgeons should be aware of risk indicators of bleeding such as older age, male sex, absence of diabetes mellitus and presence of dyslipidemia in patients with three bypassed vessels before CABG. We also recommend statisticians to utilize the decision tree model instead of logistic regression analysis in classification of risk groups.

PDF

___

1. Karthik S, Grayson AD, McCarron EE, Pullan DM, Desmond MJ. Reexploration for bleeding after coronary artery bypass surgery: risk factors, outcomes, and the effect of time delay. Ann Thorac Surg 2004;78:527-34.
2. Choong CK, Gerrard C, Goldsmith KA, Dunningham H, Vuylsteke A. Delayed re-exploration for bleeding after coronary artery bypass surgery results in adverse outcomes. Eur J Cardiothorac Surg 2007;31:834-8.
3. Dacey LJ, Munoz JJ, Baribeau YR, Johnson ER, Lahey SJ, Leavitt BJ, et al. Reexploration for hemorrhage following coronary artery bypass grafting: incidence and risk factors. Northern New England Cardiovascular Disease Study Group. Arch Surg 1998;133:442-7.
4. Al-Fayes M, Allaham A, Shawabkeh Z, Al-Naser Y, Edwan H, Abu Anzeh R. Reopening for bleeding after adult cardiac surgery. Journal of the Royal Medical Services 2011;18:67-71.
5. Han J, Kamber M. Data mining: concepts and techniques. 2nd ed. San Francisco: Morgan Kauffman; 2006.
6. Samanta B, Bird GL, Kuijpers M, Zimmerman RA, Jarvik GP, Wernovsky G, et al. Prediction of periventricular leukomalacia. Part I: Selection of hemodynamic features using logistic regression and decision tree algorithms. Artif Intell Med 2009;46:201-15. doi: 10.1016/j.artmed.2008.12.005.
7. Karaolis MA, Moutiris JA, Hadjipanayi D, Pattichis CS. Assessment of the risk factors of coronary heart events based on data mining with decision trees. IEEE Trans Inf Technol Biomed 2010;14:559-66. doi: 10.1109/TITB.2009.2038906.
8. Xiao-Bai L. A scalable decision tree system and its application in pattern recognition and intrusion detection. Decis Support Syst 2005;41:112-30.
9. Bakır B, Batmaz I, Güntürkün FA, İpekçi I, Köksal G, Özdemirel N. Defect cause modeling with decision tree and regression analysis. World Acad Sci Eng Technoly 2006;24:1-4.
10. Aitkenhead M. A co-evolving decision tree classification method. Expert Syst Appl 2008;34:18-25.
11. Mitchell TM. Machine learning. 2nd ed. New York: McGraw- Hill; 1997.
12. Kurt I, Ture M, Kurum AT. Comparing performances of logistic regression, classificationand regression tree, and neural networks for predicting coronary artery disease. Expert Syst Appl 2008;34:366-74.
13. King G, Zeng L, Explaining rare events in international relations. International Organization 2001;55:693-715.
14. King G, Zeng L. Logistic regression in rare events data. Ipolitical Analysis 2000;9:137-63.
15. Kirklin JW, Barratt-Boyes BG. Cardiac surgery. New York: John Wiley & Sons; 1986. p. 158-9.
16. Hosmer DW, Lemeshow S. Applied logistic regression. 2nd ed. New York: John Wiley & Sons; 2000.
17. Rudolfer SM, Paliouras G, Peers IS. A comparison of logistic regression to decision tree induction in the diagnosis of carpal tunnel syndrome. Comput Biomed Res 1999;32:391-414.
18. Jenhani I, Ben N, Elouedi Z. Decision trees as possibilistic classifiers. Int J Approx Reason 2008;48:784-807.
19. Detsky AS, Naglie G, Krahn MD, Redelmeier DA, Naimark D. Primer on medical decision analysis: Part 2--Building a tree. Med Decis Making 1997;17:126-35.
20. Sledjeski EM, Dierker LC, Brigham R, Breslin E. The use of risk assessment to predict recurrent maltreatment: a Classification and Regression Tree Analysis (CART). Prev Sci 2008;9:28-37. doi: 10.1007/s11121-007-0079-0.
21. Agresti A. An Introduction to categorical data analysis. 2nd ed. New York: John Wiley & Sons; 2007.
22. Mehta RH, Sheng S, OBrien SM, Grover FL, Gammie JS, Ferguson TB, et al. Reoperation for bleeding in patients undergoing coronary artery bypass surgery: incidence, risk factors, time trends, and outcomes. Circ Cardiovasc Qual Outcomes 2009;2:583-90. doi: 10.1161/ CIRCOUTCOMES.109.858811.
23. McPhee S, Papadakis MA, Rabow MW. Current medical diagnosis and treatment. Fifty-first edition. New York: McGraw-Hill Medical; 2012.
24. Dao TK, Chu D, Springer J, Hiatt E, Nguyen Q. Depression and geographic status as predictors for coronary artery bypass surgery outcomes. J Rural Health 2010;26:36-43. doi: 10.1111/j.1748-0361.2009.00263.x.
25. Sangjae L. Using data envelopment analysis and decision trees for efficiency analysis and recommendation of B2C controls. Decis Support Syst 2010;49:486-97.
26. Skinner KR, Montgomery DC, Runger GC, Fowler JW, McCarville DR, Rhoads T, et al. Multivariate statistical methods for modeling and analysis of wafer probe test data. IEEE Transactions on Semiconductor Manufacturing 2002; 15:523-30.
27. Dreiseitl S, Ohno-Machado L. Logistic regression and artificial neural network classification models: a methodology review. J Biomed Inform 2002;35:352-9.
28. Rousseeuwa PJ, Christmannb A. Robustness against separation and outliers in logistic regression. Comput Stat Data Anal 2003;43:315-32.
29. Zhu M, Philpotts D, Sparks R, Stevenson M. A Hybrid approach to combining CART and logistic regression for stock ranking. Journal of Portfolio Management 2011;38:100-9.
30. Hodge VJ, Austin J. A survey of outlier detection methodologies. Artif Intell Rev 2004;22:85-126.
31. Twala BETH, Jones MC, Hand DJ. Good methods for coping with missing data in decision trees. Pattern Recognit Lett 2008;29:950-6.
32. Das U, Maiti T, Pradhan V. Bias correction in logistic regression with missing categorical covariates. J Stat Plan Inference 2010;140:2478-85.
33. Mirta B, Natasa S, Marijana ZS. Modelling small-business credit scoring by using logistic regression, neural networks and decision trees. Intell Sys Acc Fin Mgmt 2005;13:133-50.