Database Analysis of HIV Drug Resistance Patterns
Amarendra Kumar Das, Stanford University
Despite the great success of highly active antiretroviral therapy (HAART) in reducing morbidity and mortality associated with HIV, drug resistance remains a major obstacle to successful treatment, and major gaps exist in what is known about HIV drug resistance for the complex drug combinations that represent the standard of care today. There has been a rapid growth of data on HIV drug resistance now accumulating in databases such as the Stanford HIV Drug Resistance Database (HIVdb), but investigators urgently require computational tools that can make scientific sense out of such complex, time-oriented data. In our proposed IDEA application, we provide focused efforts to address this challenge through the application of a reusable, extendable software program to query, aggregate, and abstract time-oriented data in the HIVdb. Currently available tools do not allow the specification of high-level patterns for data abstraction--such as definitions of treatment outcomes--that investigators frequently utilize in data analysis, and result in time-consuming efforts for researchers to model and reuse such knowledge. By developing and validating a novel tool that formally captures clinical knowledge, we proposed to remove a major barrier to the use of longitudinal research data on HIV care. Our research plan focuses on defining a HIV drug resistance knowledgebase, extending a formal computational method for clinical data abstraction for statistical analysis, and testing the expressivity and accuracy of our approach. The results of our proposed research efforts can help investigators and providers advance knowledge of which drug regimens are best suited for individuals living with HIV infection.