在全球人口中,为糖尿病风险预测而进行少量转移学习
Shrinit Babel1, Sunit Babel2, John Hodgson3,4
1Morsani College of Medicine, University of South Florida, Tampa, FL 33602, USA.
Medicina (Kaunas, Lithuania)
|January 28, 2026
概括
少数射击域适应改善了跨多种人群的2型糖尿病 (T2DM) 风险预测. 这种方法可以在资源有限的环境中使用最小的本地数据来加强选.
科学领域:
- 计算生物学是一种计算生物学.
- 机器学习 机器学习
- 公共卫生 公共卫生
背景情况:
- 2型糖尿病 (T2DM) 影响全球超过5.37亿成年人,对低收入和中等收入国家产生重大影响.
- 由于数据和测量实践的差异,T2DM的预测模型往往无法在不同人群中进行概括.
- 许多地区的诊断资源有限,阻碍了有效的T2DM查和管理.
研究的目的:
- 评估一些射击领域适应框架,以改善跨多种人群的T2DM风险预测.
- 评估微调模型的有效性,使用小比例的本地数据进行跨人口通用化.
- 在T2DM预测中调查特定人群特征的重要性和决策值.
主要方法:
- 使用四种临床特征 (年龄,BMI,MAP,血葡萄糖) 的多层感知子,采用了几次射击域适应框架.
- 该框架在三个成年人群中进行了测试:孟加拉国,伊拉克和皮马印第安人的数据集.
- 模型在源队列上进行了预训练,并在标记目标数据的1-20%上进行了微调,与值调整相比,性能较好.
主要成果:
- 在只有源代码模型失败的情况下,Few-shot微调恢复了预测能力,显著改善了F1分数.
- 在中等性能转移中,调整提高了AUROC高达0.14个百分点,精度提高了17个百分点.
- SHAP分析揭示了特征重要性 (例如,Pima中的BMI) 和葡萄糖的决策值的特定人群变化.
结论:
- 短暂的域调整,只使用5-20%的本地标签,有效地提高了跨人群T2DM风险预测.
- 该方法利用常规可用的临床特征,并为预防性查提供可扩展,可解释的解决方案.
- 这种方法可以在多样化和资源有限的环境中民主化T2DM选.
更多相关视频
07:13Comparison of Predictive Performance of Three Lymph Node Staging Systems in Colorectal Signet Ring Cell Carcinoma Based on Machine Learning Model
Published on: April 18, 2025
509
20:36Predicting the Effectiveness of Population Replacement Strategy Using Mathematical Modeling
Published on: July 4, 2007
9.1K
相关概念视频
Conservation of Small Populations
17.1K
Small population sizes put a species at extreme risk of extinction due to a lack of variation, and a consequent decrease in adaptability. This weakens the chances of survival under pressures such as climate change, competition from other species, or new diseases. Large populations are more likely to survive pressures such as these, as such populations are more likely to harbor individuals that have genetic variants that are adaptive under new stresses. Small populations are much less...
17.1K
Global Climate Change
28.9K
Throughout its ~4.5 billion year history, the Earth has experienced periods of warming and cooling. However, the current drastic increase in global temperatures is well outside of the Earth’s cyclic norms, and evidence for human-caused global climate change is compelling. Paleoclimatology, the study of ancient climate conditions, provides ample evidence for human-caused global climate change by comparing recent conditions with those in the past.
28.9K
Predicting Molecular Geometry
45.7K
VSEPR Theory for Determination of Electron Pair Geometries
45.7K
What is Population Genetics?
64.7K
A population is composed of members of the same species that simultaneously live and interact in the same area. When individuals in a population breed, they pass down their genes to their offspring. Many of these genes are polymorphic, meaning that they occur in multiple variants. Such variations of a gene are referred to as alleles. The collective set of all the alleles within a population is known as the gene pool.
64.7K
Population Growth
28.4K
Population size is dynamic, increasing with birth rates and immigration, and decreasing with death rates and emigration. In ideal conditions with unlimited resources, populations can increase exponentially, which plots as a J-shaped growth rate curve of population size against time. This type of curve is characteristic of newly-introduced invasive species, or populations that have suffered catastrophic declines and are rebounding.
28.4K
Relative Risk
2.1K
Relative risk (RR) is a statistical measure commonly used in epidemiology to compare the likelihood of a particular event occurring between two groups. This metric is important for evaluating the relationship between exposure to a specific risk factor and the probability of a particular outcome. It plays a crucial role in medical research, public health studies, and risk assessment. Relative risk quantifies how much more (or less) likely an event is to occur in an exposed group compared to an...
2.1K
