亚型
生物标志物
维数(图论)
计算机科学
计算生物学
代谢组学
尿
选择(遗传算法)
疾病
代谢性疾病
生物信息学
医学
生物
机器学习
病理
内科学
数学
遗传学
纯数学
程序设计语言
作者
Man Zhang,Fangying Shi,Yijie Chen,Chenyu Yang,Xiangmin Zhang,Chunhui Deng,Nianrong Sun
出处
期刊:Small
[Wiley]
日期:2024-03-26
卷期号:20 (33)
被引量:1
标识
DOI:10.1002/smll.202400941
摘要
Multidimensional metabolic analysis has become a new trend in establishing efficient disease monitoring systems, as the constraints associated with relying solely on a single dimension in refined monitoring are increasingly pronounced. Here, coordination polymers are employed as derivative precursors to create multishell hollow hybrids, developing an integrated metabolic monitoring system. Briefly, metabolic fingerprints are extracted from hundreds of serum samples and urine samples, encompassing not only membranous nephropathy but also related diseases, using high-throughput mass spectrometry. With optimized algorithm and initial feature selection, the established combined panel demonstrates enhanced accuracy in both subtype differentiation (over 98.1%) and prognostic monitoring (over 95.6%), even during double blind test. This surpasses the serum biomarker panel (≈90.7% for subtyping, ≈89.7% for prognosis) and urine biomarker panel (≈94.4% for subtyping, ≈76.5% for prognosis). Moreover, after attempting to further refine the marker panel, the blind test maintains equal sensitivity, specificity, and accuracy, showcasing a comprehensive improvement over the single-fluid approach. This underscores the remarkable effectiveness and superiority of the integrated strategy in discriminating between MN and other groups. This work has the potential to significantly advance diagnostic medicine, leading to the establishment of more effective strategies for patient management.
科研通智能强力驱动
Strongly Powered by AbleSci AI