机器学习在端粒生物学障碍的诊断工作中的应用
Erika Massaccesi1, Luca Arcuri1, Giacomo Cavalca2,3
1Hematology Unit IRCCS Istituto Giannina Gaslini Genoa Italy.
HemaSphere
|January 14, 2026
概括
机器学习模型可以帮助在患有持续性细胞衰减的患者中诊断端粒生物学障碍 (TBD). 这些分析准确地重新分配未定义的病例,改善罕见疾病的诊断.
科学领域:
- 血液学 血液学 血液学
- 遗传学 是一个遗传学.
- 计算生物学 计算生物学
背景情况:
- 持续的细胞衰减和暗示端粒生物学障碍 (TBDs) 的特征提出了诊断挑战.
- 准确的结核病诊断对于有效的患者管理和治疗至关重要.
- 机器学习 (ML) 提供了分析复杂患者数据以帮助诊断的潜力.
研究的目的:
- 将监督和无监督的机器学习应用于患有持续细胞衰减和可疑结核病的患者队列.
- 在未明确诊断的患者中识别潜在的TBD患者.
- 评估ML在重新分配诊断和改善罕见疾病的诊断工作中的实用性.
主要方法:
- 分析了140名患有持续性细胞衰减或可疑结核病的患者队列.
- 监督随机森林模型训练47名已确诊的患者 (结核病与其他诊断).
- 无监督集群适用于整个队列,以确定不同的患者群体.
主要成果:
- 监督分析预测了17.2%的未确定的诊断 (UD) 患者是潜在的TBD.
- 无监督分析确定了4个与分子诊断有显著关联的集群;TBD患者在集群1和2中占主导地位.
- 端粒长度和粘膜皮肤异常是结核病和其他诊断群体之间的关键区分因素.
结论:
- 机器学习模型可以有效地重新分配大量未定义或错误分类的TBD病例.
- 开发的ML方法在改善TBD等罕见和具有挑战性的疾病的诊断工作方面显示出希望.
- 将ML集成到诊断途径中可以提高血液学疾病识别的准确性和效率.
相关概念视频
Telomeres and Telomerase
26.8K
In eukaryotic DNA replication, a single-stranded DNA fragment remains at the end of a chromosome after the removal of the final primer. This section of DNA cannot be replicated in the same manner as the rest of the strand because there is no 3’ end to which the newly synthesized DNA can attach. This non-replicated fragment results in gradual loss of the chromosomal DNA during each cell duplication. Additionally, it can induce a DNA damage response by enzymes that recognize single-stranded...
26.8K
Telomeres and Telomerase
7.0K
7.0K
Replicative Cell Senescence
4.3K
Replicative cell senescence is a property of cells that allows them to divide a finite number of times throughout the organism's lifespan while preventing excessive proliferation. Replicative senescence is associated with the gradual loss of the telomere — short, repetitive DNA sequences found at the end of the chromosomes. Telomeres are bound by a group of proteins to form a protective cap on the ends of chromosomes. Embryonic stem cells express telomerase — an enzyme that adds...
4.3K


