Türk Pediatrik Popülasyonunda Makine Öğrenimi Modelleri Kullanılarak LDL-K Tahmini ve Doğrudan Ölçülen ve Hesaplanan LDL-K ile Karşılaştırılması

Amaç: Çocuklarda lipid profillerinin değerlendirilmesi, dislipideminin erken saptanması için kritik öneme sahiptir. Düşük yoğunluklu lipoprotein kolesterol (LDL-K), dislipidemik hastaların teşhis ve tedavisinde en sık kullanılan ölçümlerden biridir. Bu nedenle, LDL-K düzeylerinin doğru belirlenmesi, lipid anormalliklerinin yönetimi için kritik öneme sahiptir. Bu çalışmada, Türk pediatrik popülasyonunda çeşitli LDL-K tahmin formüllerini güçlü makine öğrenmesi algoritmalarıyla karşılaştırmayı amaçladık. Gereç ve Yöntemler: Bu çalışmaya Sivas Cumhuriyet Üniversitesi Hastanesi'nde tedavi gören 18 yaş altı 2,563 çocuk dahil edildi. LDL-K değerleri Roche direkt yöntemi kullanılarak ölçüldü ve Friedewald, Martin/Hopkins, Chen, Anandaraja ve Hattori formülleri ile makine öğrenmesi modelleri (Ridge, Lasso, elastic net, destek vektör regresyonu, rastgele orman, gradyan artırma ve aşırı gradyan artırma) kullanılarak tahmin edildi. Tahminler ve direkt ölçümler arasındaki uyum hem genel olarak hem de LDL-K ve TG alt seviyeleri için ayrı ayrı değerlendirildi. Ayrıca, doğrusal regresyon analizleri gerçekleştirilmiş olup her bir LDL-K tahmini ile direkt ölçüm yöntemi arasındaki fark artık hata grafikleri ile gösterilmiştir. Bulgular: Tahminlenen LDL-K değerleri ile Roche direkt metodu ile ölçülen LDL-K değerleri arasındaki uyum, makine öğrenmesi modelleri için yaklaşık yüzde 0,92-0,93 ve LDL-K tahmin formülleri için yaklaşık %0,85 idi. Destek vektör regresyonu en uyumlu sonuçları (uyum=0,938) verirken, Hattori ve Martin-Hopkins formülleri en az uyumlu sonuçları (uyum=0,851) vermiştir. Sonuç: Makine öğrenmesi modelleri, LDL-K tahmin formüllerine kıyasla daha uyumlu LDL-K tahmini yaptığından, makine öğrenmesi modelleri, geleneksel LDL-K tahmin formülleri ve doğrudan analizlerin yerine kullanılabilir.

Anahtar Kelimeler:

Kardiyovasküler Hastalıklar, Kolesterol, Lipoproteinler, Düşük-Yoğunluklu Lipoprotein, Makine Öğrenimi

Estimation of LDL-C using machine learning models and its comparison with directly measured and calculated LDL-C in Turkish pediatric population

Objective: The assessment of lipid profiles in children is critical for the early detection of dyslipidemia. Low-density lipoprotein cholesterol (LDL-C) is one of the most often used measures in diagnosing and treating patients with dyslipidemia. Therefore, accurate determination of LDL-C levels is critical for managing lipid abnormalities. In this study, we aimed to compare various LDL-C estimating formulas with powerful machine-learning (ML) algorithms in a Turkish pediatric population. Materials and Methods: This study included 2,563 children under 18 who were treated at Sivas Cumhuriyet University Hospital in Sivas, Türkiye. LDL-C was measured directly using Roche direct assay and estimated using Friedewald's, Martin/Hopkins', Chen's, Anandaraja's, and Hattori's formulas, as well as ML predictive models (i.e., Ridge, Lasso, elastic net, support vector regression, random forest, gradient boosting and extreme gradient boosting). The concordances between the estimates and direct measurements were assessed overall and separately for the LDL-C and TG sublevels. Linear regression analyses were also carried out, and residual error plots were created between each LDL-C estimation and direct measurement method. Results: The concordance was approximately 0.92-0.93 percent for ML models, and around 0.85 percent for LDL-C estimating formulas. The SVR formula generated the most concordant results (concordance=0.938), while the Hattori and Martin-Hopkins formulas produced the least concordant results (concordance=0.851). Conclusion: Since ML models produced more concordant LDL-C estimates compared to LDL-C estimating formulas, ML models can be used in place of traditional LDL-C estimating formulas and direct assays.

Keywords:

Cardiovascular Diseases, Cholesterol, Lipoproteins, Low-Density Lipoprotein, Machine Learning,

PDF

___

Berenson GS, Srinivasan SR, Bao W, Newman W, Tracy RE, Wattigney WA, et al. Association between multiple cardiovascular risk factors and atherosclerosis in children and young adults. The Bogalusa Heart Study. N Engl J Med. 1998; 338 (23): 1650-1656. doi: 10.1056/NEJM199806043382302.
Daniels SR, Greer FR. Lipid screening and cardiovascular health in childhood. Pediatrics. 2008; 122 (1): 198-208. doi: 10.1542/peds.2008-1349.
McGill HC, McMahan CA, Zieske AW, Malcom GT, Tracy RE, Strong JP. Effects of nonlipid risk factors on atherosclerosis in youth with a favorable lipoprotein profile. Circulation. 2001; 103 (11): 1546-1550. doi: 10.1161/01.cir.103.11.1546.
Gidding SS. Cholesterol Guidelines Debate. Pediatrics. 2001; 107 (5): 1229–1230. https://doi.org/10.1542/peds.107.5.1229.
Fox KM, Wang L, Gandra SR, Quek RGW, Li L, Baser O. Clinical and economic burden associated with cardiovascular events among patients with hyperlipidemia: A retrospective cohort study. BMC Cardiovasc Disord. 2016; 16 (1). doi: 10.1186/s12872-016-0190-x.
Molavi F, Namazi N, Asadi M, Sanjari M, Motlagh ME, Shafiee G, et al. Comparison common equations for LDL-C calculation with direct assay and developing a novel formula in Iranian children and adolescents: The CASPIAN v study. Lipids Health Dis. 2020; 19 (1). https://doi.org/10.1186/s12944-020-01306-7.
Silverman MG, Ference BA, Im K, Wiviott SD, Giugliano RP, Grundy SM, et al. Association between lowering LDL-C and cardiovascular risk reduction among different therapeutic interventions: A systematic review and meta-analysis. JAMA - J Am Med Assoc. 2016; 316 (12): 1289-1297. doi: 10.1001/jama.2016.13985.
CDC. Centers for Disease Control and Prevention National Reference System for Cholesterol - Cholesterol Reference Method Laboratory Network - Total Cholesterol - Certification Protocol for Manufacturers-Revised. 2004;(cited 2022 April 20). Available from: http://www.cdc.gov/labstandards/pdf/crmln/FrozVsFreshProtocolOct04.pdf
Friedewald Wt Fau - Levy RI, Levy Ri Fau - Fredrickson DS, Fredrickson DS, Clin C. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem. 1972 Jun; 18 (6): 499-502.
De Cordova CMM, Schneider CR, Juttel ID, De Cordova MM. Comparison of LDL-cholesterol direct measurement with the estimate using the Friedewald formula in a sample of 10,664 patients. Arq Bras Cardiol. 2004; 83 (6). doi: 10.1590/s0066-782x2004001800006.
de Cordova CMM, de Cordova MM. A new accurate, simple formula for LDL-cholesterol estimation based on directly measured blood lipids from a large cohort. Ann Clin Biochem. 2013; 50 (1): 13-9. doi: 10.1258/acb.2012.011259.
Türkalp I, Çil Z, Özkazanç D. Analytical performance of a direct assay for LDL-cholesterol: A comparative assessment versus Friedewald’s formula. Anadolu Kardiyol Derg. 2005; 5 (1):13-17. https://pubmed.ncbi.nlm.nih.gov/15755695/.
Hermans MP, Ahn SA, Rousseau MF. Novel unbiased equations to calculate triglyceride-rich lipoprotein cholesterol from routine non-fasting lipids. Cardiovasc Diabetol. 2014; 13 (1). https://doi.org/10.1186/1475-2840-13-56.
Martin SS, Blaha MJ, Elshazly MB, Toth PP, Kwiterovich PO, Blumenthal RS, et al. Comparison of a novel method vs the Friedewald equation for estimating low-density lipoprotein cholesterol levels from the standard lipid profile. JAMA - J Am Med Assoc. 2013; 310 (19): 2061-2068. doi: 10.1001/jama.2013.280532.
Grundy SM, Stone NJ, Bailey AL, Beam C, Birtcher KK, Blumenthal RS, et al. 2018 AHA/ACC/AACVPR/AAPA/ABC/ACPM/ADA/AGS/APhA/ASPC/NLA/PCNA Guideline on the Management of Blood Cholesterol: A Report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines. Vol. 139, Circulation. 2019.
Chen Y, Zhang X, Pan B, Jin X, Yao H, Chen B, et al. A modified formula for calculating low-density lipoprotein cholesterol values. Lipids Health Dis. 2010; 9 (1). doi: 10.1186/1476-511X-9-52.
Anandaraja S, Narang R, Godeswar R, Laksmy R, Talwar KK. Low-density lipoprotein cholesterol estimation by a new formula in Indian population. Int J Cardiol. 2005; 102 (1): 117-120. doi: 10.1016/j.ijcard.2004.05.009.
Hattori Y, Suzuki M, Tsushima M, Yoshida M, Tokunaga Y, Wang Y, et al. Development of approximate formula for LDL-chol, LDL-apo B and LDL- chol/LDL-apo B as indices of hyperapobetalipoproteinemia and small dense LDL. Atherosclerosis. 1998; 138 (2): 289-299. doi: 10.1016/s0021-9150(98)00034-3.
Anudeep PP, Kumari S, Rajasimman AS, Nayak S, Priyadarsini P. Machine learning predictive models of LDL-C in the population of eastern India and its comparison with directly measured and calculated LDL-C. Ann Clin Biochem. 2022; 59 (1): 76-86. doi: 10.1177/00045632211046805.
Çubukçu HC, Topcu Dİ. Estimation of Low-Density Lipoprotein Cholesterol Concentration Using Machine Learning. Lab Med. 2022; 53 (2): 161-171. doi: 10.1093/labmed/lmab065.
Tsigalou C, Panopoulou M, Papadopoulos C, Karvelas A, Tsairidis D, Anagnostopoulos K. Estimation of low-density lipoprotein cholesterol by machine learning methods. Clin Chim Acta. 2021; 517: 108-116. doi: 10.1016/j.cca.2021.02.020.
Kwon Y-J, Lee H, Baik SJ, Chang H-J, Lee J-W. Comparison of a Machine Learning Method and Various Equations for Estimating Low-Density Lipoprotein Cholesterol in Korean Populations. Front Cardiovasc Med 2022 Feb 10; 9(February): 824574. doi: 10.3389/fcvm.2022.824574.
Hastie, Trevor, Tibshirani, Robert, Friedman J. The Elements of Statistical Learning The Elements of Statistical LearningData Mining, Inference, and Prediction, Second Edition. Springer series in statistics. 2009.
Hastie T, Tibshirani R, Wainwright M. Statistical learning with sparsity: The lasso and generalizations. Statistical Learning with Sparsity: The Lasso and Generalizations. 2015.
Kuhn M. Building predictive models in R using the caret package. J Stat Softw. 2008; 28 (5): doi: 10.18637/jss.v028.i05.
Kuhn, Max. Applied Predictive Modeling. Springer. 2013.
Roth GA, Mensah GA, Johnson CO, Addolorato G, Ammirati E, Baddour LM, et al. Global Burden of Cardiovascular Diseases and Risk Factors, 1990-2019: Update From the GBD 2019 Study. Vol. 76, Journal of the American College of Cardiology. 2020.
Lim JS, Kim EY, Kim JH, Yoo JH, Yi KH, Chae HW, et al. 2017 clinical practice guidelines for dyslipidemia of korean children and adolescents. Ann Pediatr Endocrinol Metab. 2020; 25 (4): 454-462. doi: 10.3345/cep.2020.01340.
Lund SS, Petersen M, Frandsen M, Smidt UM, Parving HH, Vaag AA, et al. Agreement between fasting and postprandial LDL cholesterol measured with 3 methods in patients with type 2 diabetes mellitus. Clin Chem. 2011; 57 (2): 298–308. https://doi.org/10.1373/clinchem.2009.133868.