Leveraging public health genotyping databases for near real-time HIV surveillance

点击翻译按钮获取中文摘要

Summary In a collaboration between Los Alamos National Laboratory, Imperial College London, Colorado Department of Public Health and Environment, and Michigan Department of Health and Human Services, we propose to develop a near real-time surveillance tool leveraging existing databases to support public health efforts in testing, treatment, and prevention. Overall, we aim to develop a computational pipeline that will use data from a public health database and function as a practical surveillance tool. The pipeline's design will be modular to allow for customization and independent updating focusing on creating actionable surveillance reports in near real-time. Our approach is based on reconstructing the underlying transmission network that generated a particular HIV phylogeny. Such methods, aka. source attribution methods, have been shown to have less error, be less sensitive to sampling artifacts, provide more actionable information about how HIV spreads among age/risk/single or multiple-source connections, and to be able to identify unsampled persons, all better than simple genetic clustering methods. Thus, correct use of inferred transmission network information would improve resource allocation, allow more accurate and therefore faster interventions, and ultimately prevent more persons from becoming infected. We will also include data quality control measures and automatic checks for robustness and repeatability of the phylodynamic inferences. To achieve this, we divide the project into two aims: 1) Develop a near real-time surveillance tool for practical public health use, and 2) Develop phylodynamic analysis methods into modules that can be used to enhance the utility of the surveillance tool. We propose several innovative scientific advancements to the current state of the art of phylodynamic methodology and include aspects of quality control, robustness, and repeatability to solve the demanding computational tasks of integrating those methods into a surveillance tool. We carefully consider ethical and legal issues that may arise. We emphasize that inferred transmission network information is sensitive data that must be handled in the same responsible way as current partner services data. We will use data from Colorado and Michigan, both states with significant HIV epidemics, but with different demographic and epidemic situations, which has the advantage of forcing us to develop a flexible and universally meaningful surveillance system.