使用条件生成对抗网络生成合成个人健康数据,并与差异隐私相结合
Chang Sun1, Johan van Soest2, Michel Dumontier1
1Institute of Data Science, Faculty of Science and Engineering, Maastricht University, Maastricht, The Netherlands; Department of Advanced Computing Sciences, Faculty of Science and Engineering, Maastricht University, Maastricht, The Netherlands.
Journal of biomedical informatics
|June 2, 2023
概括
生成现实的合成健康数据是具有挑战性的. 一个新的差别私有条件生成对抗网络 (DP-CGANS) 模型解决了隐私和少数类数据挑战,改善了数据实用性和隐私平衡.
科学领域:
- 医疗信息学 医疗信息学
- 数据科学数据科学数据科学
- 保护隐私的技术 保护隐私的技术
背景情况:
- 个人健康数据是有价值的,但由于隐私和法律约束,无法访问.
- 合成数据提供了一个有希望的解决方案,但现实主义,隐私保护和处理不平衡的数据集仍然存在挑战.
- 现有的方法在少数类数据模拟和捕获变量依赖性方面遇到了困难.
研究的目的:
- 提出一种新的差异私有有条件生成对抗网络 (DP-CGANS) 模型,用于生成现实的,保护隐私的合成个人健康数据.
- 解决综合健康数据生成的具体挑战,包括少数阶级的代表性和变量之间的依赖性.
- 确保数据实用性和患者隐私之间的平衡.
主要方法:
- 开发了一个DP-CGANS模型,涉及数据转换,采样,调节和网络培训.
- 分别将分类变量和连续变量分别转换为潜空间.
- 整合了一个条件向量来表示少数阶级,并将噪音注入梯度以实现差异性隐私.
主要成果:
- 与最先进的模型相比,DP-CGANS模型在捕捉变量依赖性方面表现出卓越的性能.
- 对社会经济和现实世界健康数据集的评估显示出强烈的统计相似性,机器学习实用性和隐私保护.
- 该模型有效地平衡了数据实用性和隐私,即使有不平衡的类,异常分布和数据稀疏性.
结论:
- DP-CGANS是一种有效的方法,用于生成高保真性,保护隐私的合成个人健康数据.
- 该模型成功地克服了合成健康数据生成的关键挑战,特别是不平衡和复杂的数据集.
- 这种方法可以提高研究价值的健康数据的可访问性,同时保持严格的隐私标准.
相关概念视频
Censoring Survival Data
154
Survival analysis is a statistical method used to analyze time-to-event data, often employed in fields such as medicine, engineering, and social sciences. One of the key challenges in survival analysis is dealing with incomplete data, a phenomenon known as "censoring." Censoring occurs when the event of interest (such as death, relapse, or system failure) has not occurred for some individuals by the end of the study period or is otherwise unobservable, and it might have many different...
154
Genome-wide Association Studies-GWAS
13.7K
Genome-wide association studies or GWAS are used to identify whether common SNPs are associated with certain diseases. Suppose specific SNPs are more frequently observed in individuals with a particular disease than those without the disease. In that case, those SNPs are said to be associated with the disease. Chi-square analysis is performed to check the probability of the allele likely to be associated with the disease.
GWAS does not require the identification of the target gene involved in...
GWAS does not require the identification of the target gene involved in...
13.7K
Genetic Material
1.9K
Within the human body, a complex and detailed system of trillions of cells works in unison to sustain life. Each cell houses a nucleus, which contains 46 chromosomes divided into 23 pairs. Chromosomes are highly coiled structures made of the genetic material DNA. These chromosomes are essential carriers of genetic information, with half inherited from the mother through her egg and the other half from the father's sperm, combining to create the unique genetic makeup of an individual.
1.9K
What is Genetic Engineering?
74.4K
Overview
74.4K
Model Approaches for Pharmacokinetic Data: Distributed Parameter Models
101
Pharmacokinetic models are mathematical constructs that represent and predict the time course of drug concentrations in the body, providing meaningful pharmacokinetic parameters. These models are categorized into compartment, physiological, and distributed parameter models.
The distributed parameter models are specifically designed to account for variations and differences in some drug classes. This model is particularly useful for assessing regional concentrations of anticancer or...
The distributed parameter models are specifically designed to account for variations and differences in some drug classes. This model is particularly useful for assessing regional concentrations of anticancer or...
101


