常规公式与机器学习模型估计血清低密度脂蛋白胆固醇的比较
Dhuhli, R.A., Riyami, N.A., Farsi, F.A., Khamyasi, A.A., Saadi, H.A., Alawi, S.A., Alfarai, Q., Zidoum, H., Alsalmani, F., Almarshodi, Y., & Zadjali, F. (2026). Comparison of conventional formulas and machine learning models for estimating serum low-density lipoprotein cholesterol. Practical Laboratory Medicine. https://doi.org/10.1016/j.plabm.2026.e00524
Dhuhli RA, Riyami NA, Farsi FA, Khamyasi AA, Saadi HA, Alawi SA, et al. Comparison of conventional formulas and machine learning models for estimating serum low-density lipoprotein cholesterol. Practical Laboratory Medicine. 2026; doi: 10.1016/j.plabm.2026.e00524
Dhuhli RA, Riyami NA, Farsi FA, et al. Comparison of conventional formulas and machine learning models for estimating serum low-density lipoprotein cholesterol[J]. Practical Laboratory Medicine. 2026. DOI: 10.1016/j.plabm.2026.e00524.
@article{dhuhli2026,
author = {Rashid Al Dhuhli and Noureldin Al Riyami and Fatma Al Farsi and Ahmed Al Khamyasi and Hamza Al Saadi and Said Al Alawi and Qasim Alfarai and Hasan Zidoum and Fatma Alsalmani and Yaqoot Almarshodi and Fatma Zadjali},
title = {Comparison of conventional formulas and machine learning models for estimating serum low-density lipoprotein cholesterol},
journal = {Practical Laboratory Medicine},
year = {2026},
doi = {10.1016/j.plabm.2026.e00524},
note = {PMID: 41853761},
}
TY - JOUR AU - Rashid Al Dhuhli AU - Noureldin Al Riyami AU - Fatma Al Farsi AU - Ahmed Al Khamyasi AU - Hamza Al Saadi AU - Said Al Alawi AU - Qasim Alfarai AU - Hasan Zidoum AU - Fatma Alsalmani AU - Yaqoot Almarshodi AU - Fatma Zadjali TI - Comparison of conventional formulas and machine learning models for estimating serum low-density lipoprotein cholesterol T2 - Practical Laboratory Medicine PY - 2026 DO - 10.1016/j.plabm.2026.e00524 AN - PMID:41853761 ER -
准确估计低密度脂蛋白胆固醇(LDL-C)对于心血管风险评估和监测至关重要。由于β定量法操作繁琐且不适用于常规使用,已开发了多种方程来估计LDL-C浓度。然而,在甘油三酯水平升高、低LDL-C或非空腹样本等临床条件下,传统方程的性能可能下降。本研究比较了常规估计公式(Friedewald、Martin-Hopkins和Sampson-Nielsen公式)与机器学习模型在估计血清LDL-C方面的性能。研究使用了一个大型临床实验室数据库,涵盖了广泛的脂质谱特征。结果显示,机器学习模型在广泛的临床条件下均优于传统方程,特别是在甘油三酯升高和低LDL-C水平的情况下。机器学习方法通过利用多种脂质参数之间的非线性关系,提供了更准确的LDL-C估计,有望改善基于脂质的心血管风险分层,特别是在传统公式准确性有限的临床场景中。