Tools to Address the Challenges of Preserving Privacy in Sharing and Analysis of Biomedical Data

项目摘要 了解一种疾病或病症需要大规模挖掘遗传、表观遗传、 环境和临床观察，从大量的人类数据。广泛和 要破译已收集到的大量数据，就必须方便地获取这些数据 在个人和群体层面。然而，保护人权与保护人权之间存在直接冲突。 患者和研究参与者的隐私，以及共享遗传、表观遗传和临床数据， 生物医学进步。这种冲突的部分原因是， 并分析个人生物数据和建立理论和实施的领域 数据隐私和安全。这个项目的目标是双重的：1）量化私人信息 以系统的方式，从各种类型的人类衍生的生物数据泄漏， 数据共享工作，以及2）为各种数据开发增强隐私的分析软件， 分子水平（基因组学、转录组学、表型数据）。我们将创造一个不断发展的 和模块化工具套件，以量化和保护隐私;该套件将有能力 采用新的数据模式和分析需求。建议的工具将有助于 防止未来灾难性的隐私泄露，这可能导致失去所有医疗服务。 可操作的数据;使所有研究人员的数据访问民主化;并在患者之间建立信任 和研究人员，从而增加参与研究。

Project Summary Understanding a single disease or condition requires large-scale mining of genetic, epigenetic, environmental, and clinical observations from large amounts of human data. Widespread and easy access to such data is imperative to decipher the vast trove of data that has been collected at the individual and population level. However, there is a direct conflict between protecting the privacy of patients and research participants and sharing genetic, epigenetic, and clinical data for biomedical advances. This conflict is partly due to a disconnect between the fields that generate and analyze personal biological data and the fields that establish theories and implementations for data privacy and security. The goal of this project is two-fold: 1) to quantify private information leakages from various types of human-derived biological data in a systematic manner to inform data-sharing efforts and 2) to develop privacy-enhancing analysis software for data at various molecular levels (genomics, transcriptomics, phenotype data) at scale. We will create an evolving and modular tool suite to both quantify and preserve privacy; this suite will have the ability to be adopted to new data modalities and analysis needs as they arise. The proposed tools will help prevent future catastrophic privacy leaks, which may result in a loss of access to all medically actionable data; democratize data access for all researchers; and create trust between patients and researchers, thus increasing participation in studies.