Comparison of risk models for mortality and cardiovascular events between machine learning and conventional logistic regression analysis

English version

Autoři: Shinya Suzuki ^aff001; Takeshi Yamashita ^aff001; Tsuyoshi Sakama ^aff002; Takuto Arita ^aff001; Naoharu Yagi ^aff001; Takayuki Otsuka ^aff001; Hiroaki Semba ^aff001; Hiroto Kano ^aff001; Shunsuke Matsuno ^aff001; Yuko Kato ^aff001; Tokuhisa Uejima ^aff001; Yuji Oikawa ^aff001; Minoru Matsuhama ^aff003; Junji Yajima ^aff001
Působiště autorů: Department of Cardiovascular Medicine, The Cardiovascular Institute, Tokyo, Japan ^aff001; Sigmaxyz, Inc, Tokyo, Japan ^aff002; Department of Cardiovascular Surgery, The Cardiovascular Institute, Tokyo, Japan ^aff003
Vyšlo v časopise: PLoS ONE 14(9)
Kategorie: Research Article
prolekare.web.journal.doi_sk: https://doi.org/10.1371/journal.pone.0221911

Souhrn

Aims

Non-linear models by machine learning may identify different risk factors with different weighting in comparison to conventional linear models.

Methods and results

The analyses were performed in 15,933 patients included in the Shinken Database (SD) 2004–2014 (n = 22,022) for whom baseline data of blood sampling and ultrasound cardiogram and follow-up data at 2 years were available. Using non-linear models with machine learning software, 118 risk factors and their weighting of risk for all-cause mortality, heart failure (HF), acute coronary syndrome (ACS), ischemic stroke (IS), and intracranial hemorrhage (ICH) were identified, where the top two risk factors were albumin/hemoglobin, left ventricular ejection fraction/history of HF, history of ACS/anti-platelet use, history of IS/deceleration time, and history of ICH/warfarin use. The areas under the curve of the developed models for each event were 0.900, 0.912, 0.879, 0.758, and 0.753, respectively.

Conclusion

Here, we described our experience with the development of models for predicting cardiovascular prognosis by machine learning. Machine learning could identify risk predicting models with good predictive capability and good discrimination of the risk impact.

Klíčová slova:

stroke – Physical sciences – Research and analysis methods – Computer and information sciences – Mathematics – Medicine and health sciences – Pathology and laboratory medicine – Diagnostic medicine – Signs and symptoms – Statistics – Mathematical and statistical techniques – Statistical methods – Pharmaceutics – Drug therapy – Neurology – Cerebrovascular diseases – Ischemic stroke – Vascular medicine – Cardiology – Blood pressure – Artificial intelligence – Machine learning – Hemorrhage – Heart failure – Ejection fraction – Antiplatelet therapy

Zdroje

1. Chen JH, Asch SM. Machine Learning and Prediction in Medicine—Beyond the Peak of Inflated Expectations. N Engl J Med. 2017;376:2507–2509. doi: 10.1056/NEJMp1702071 28657867

2. Breslow NE. Analysis of Survival Data under the Proportional Hazards Model. International Statistical Review / Revue Internationale de Statistique. 1975;43:45–57.

3. Deo RC. Machine Learning in Medicine. Circulation. 2015;132:1920–1930. doi: 10.1161/CIRCULATIONAHA.115.001593 26572668

4. Weng SF, Reps J, Kai J, Garibaldi JM, Qureshi N. Can machine-learning improve cardiovascular risk prediction using routine clinical data? PLoS One. 2017;12:e0174944. doi: 10.1371/journal.pone.0174944 28376093

5. Suzuki S, Yamashita T, Ohtsuka T, Sagara K, Uejima T, Oikawa Y, et al. Prevalence and prognosis of patients with atrial fibrillation in Japan: a prospective cohort of Shinken Database 2004. Circ J. 2008;72:914–920. doi: 10.1253/circj.72.914 18503216

6. Suzuki S, Yamashita T, Otsuka T, Sagara K, Uejima T, Oikawa Y, et al. Recent mortality of Japanese patients with atrial fibrillation in an urban city of Tokyo. J Cardiol. 2011;58:116–123. doi: 10.1016/j.jjcc.2011.06.006 21820280

7. Matsuo S, Imai E, Horio M, Yasuda Y, Tomita K, Nitta K, et al. Revised equations for estimated GFR from serum creatinine in Japan. Am J Kidney Dis. 2009;53:982–992. doi: 10.1053/j.ajkd.2008.12.034 19339088

8. DataRobot [2019/3/14]. Available from: https://www.datarobot.com/.

9. Kang J, Schwartz R, Flickinger J, Beriwal S. Machine Learning Approaches for Predicting Radiation Therapy Outcomes: A Clinician's Perspective. Int J Radiat Oncol Biol Phys. 2015;93:1127–1135. doi: 10.1016/j.ijrobp.2015.07.2286 26581149

10. Brooks Carthon JM, Jarrin O, Sloane D, Kutney-Lee A. Variations in postoperative complications according to race, ethnicity, and sex in older adults. J Am Geriatr Soc. 2013;61:1499–1507. doi: 10.1111/jgs.12419 24006851

11. Breiman L. Random Forests. Machine Learning. 2001;45:5–32.

12. Friedman J. Greedy boosting approximation: a gradient boosting machine. Ann Stat. 2001;29:1189–1232.

13. Wright J, Yang AY, Ganesh A, Sastry SS, Ma Y. Robust face recognition via sparse representation. IEEE Trans Pattern Anal Mach Intell. 2009;31:210–227. doi: 10.1109/TPAMI.2008.79 19110489

14. Wang L, Tong L, Yan B, Lei Y, Wang L, Zeng Y, et al. Sparse models for visual image reconstruction from fMRI activity. Biomed Mater Eng. 2014;24:2963–2969. doi: 10.3233/BME-141116 25227003

15. Zou H, Hastie T. Regularization and Variable Selection via the Elastic Net. Journal of the Royal Statistical Society Series B (Statistical Methodology). 2005;67:301–320.

16. Vapnik VN. An overview of statistical learning theory. IEEE Trans Neural Netw. 1999;10:988–999. doi: 10.1109/72.788640 18252602

17. Unnikrishnan P, Kumar DK, Poosapadi Arjunan S, Kumar H, Mitchell P, Kawasaki R. Development of Health Parameter Model for Risk Prediction of CVD Using SVM. Comput Math Methods Med. 2016;2016:3016245. doi: 10.1155/2016/3016245 27594895

18. Yu W, Liu T, Valdez R, Gwinn M, Khoury MJ. Application of support vector machine modeling for prediction of common diseases: the case of diabetes and pre-diabetes. BMC Med Inform Decis Mak. 2010;10:16. doi: 10.1186/1472-6947-10-16 20307319

19. Ennis M, Hinton G, Naylor D, Revow M, Tibshirani R. A comparison of statistical learning methods on the Gusto database. Stat Med. 1998;17:2501–2508. 9819841

20. Jung HY, Kim SH, Jang HM, Lee S, Kim YS, Kang SW, et al. Individualized prediction of mortality using multiple inflammatory markers in patients on dialysis. PLoS One. 2018;13:e0193511. doi: 10.1371/journal.pone.0193511 29494637

21. Gotsman I, Shauer A, Zwas DR, Tahiroglu I, Lotan C, Keren A. Low serum albumin: A significant predictor of reduced survival in patients with chronic heart failure. Clin Cardiol. 2019;42:365–372. doi: 10.1002/clc.23153 30637771

22. Nakano H, Omote K, Nagai T, Nakai M, Nishimura K, Honda Y, et al. Comparison of Mortality Prediction Models on Long-Term Mortality in Hospitalized Patients With Acute Heart Failure- The Importance of Accounting for Nutritional Status. Circ J. 2019;83:614–621. doi: 10.1253/circj.CJ-18-1243 30700666

23. Nishi I, Seo Y, Hamada-Harimura Y, Yamamoto M, Ishizu T, Sugano A, et al. Geriatric nutritional risk index predicts all-cause deaths in heart failure with preserved ejection fraction. ESC Heart Fail. 2019; doi: 10.1002/ehf2.12405 30706996

24. Xia M, Zhang C, Gu J, Chen J, Wang LC, Lu Y, et al. Impact of serum albumin levels on long-term all-cause, cardiovascular, and cardiac mortality in patients with first-onset acute myocardial infarction. Clin Chim Acta. 2018;477:89–93. doi: 10.1016/j.cca.2017.12.014 29241048

25. Yang LJ, Feng YX, Li T, Jiao YR, Yao HC, Zhang DY. Serum albumin levels might be an adverse predictor of long term mortality in patients with acute myocardial infarction. Int J Cardiol. 2016;223:647–648. doi: 10.1016/j.ijcard.2016.08.251 27567231

26. Plakht Y, Gilutz H, Shiyovich A. Decreased admission serum albumin level is an independent predictor of long-term mortality in hospital survivors of acute myocardial infarction. Soroka Acute Myocardial Infarction II (SAMI-II) project. Int J Cardiol. 2016;219:20–24. doi: 10.1016/j.ijcard.2016.05.067 27257851

27. Ma L, Zhao S. Risk factors for mortality in patients undergoing hemodialysis: A systematic review and meta-analysis. Int J Cardiol. 2017;238:151–158. doi: 10.1016/j.ijcard.2017.02.095 28341375

28. Chen CW, Drechsler C, Suntharalingam P, Karumanchi SA, Wanner C, Berg AH. High Glycated Albumin and Mortality in Persons with Diabetes Mellitus on Hemodialysis. Clin Chem. 2017;63:477–485. doi: 10.1373/clinchem.2016.258319 27737895

29. Eriguchi R, Obi Y, Streja E, Tortorici AR, Rhee CM, Soohoo M, et al. Longitudinal Associations among Renal Urea Clearance-Corrected Normalized Protein Catabolic Rate, Serum Albumin, and Mortality in Patients on Hemodialysis. Clin J Am Soc Nephrol. 2017;12:1109–1117. doi: 10.2215/CJN.13141216 28490436

30. Kannel WB, McGee D, Gordon T. A general cardiovascular risk profile: the Framingham Study. Am J Cardiol. 1976;38:46–51. doi: 10.1016/0002-9149(76)90061-8 132862

31. Anderson KM, Wilson PW, Odell PM, Kannel WB. An updated coronary risk profile. A statement for health professionals. Circulation. 1991;83:356–362. doi: 10.1161/01.cir.83.1.356 1984895

32. Zimmerman JE, Kramer AA, McNair DS, Malila FM. Acute Physiology and Chronic Health Evaluation (APACHE) IV: hospital mortality assessment for today's critically ill patients. Crit Care Med. 2006;34:1297–1310. doi: 10.1097/01.CCM.0000215112.84523.F0 16540951

33. Miro O, Rossello X, Gil V, Martin-Sanchez FJ, Llorens P, Herrero-Puente P, et al. Predicting 30-Day Mortality for Patients With Acute Heart Failure in the Emergency Department: A Cohort Study. Ann Intern Med. 2017;167:698–705. doi: 10.7326/M16-2726 28973663

34. Win S, Hussain I, Hebl VB, Dunlay SM, Redfield MM. Inpatient Mortality Risk Scores and Postdischarge Events in Hospitalized Heart Failure Patients: A Community-Based Study. Circ Heart Fail. 2017;10.

35. Chichareon P, Modolo R, van Klaveren D, Takahashi K, Kogame N, Chang CC, et al. Predictive ability of ACEF and ACEF II score in patients undergoing percutaneous coronary intervention in the GLOBAL LEADERS study. Int J Cardiol. 2019; doi: 10.1016/j.ijcard.2019.02.043 30846254