Optimizing clinical trials with Open Trial Chain

Abstract

The objective of this thesis is to study the challenges of data sharing in healthcare (namely clinical trials), and propose the use of Open Algorithms (OPAL) as a viable solution for research collaboration that allows for access to data without compromising data ownership (data is only used once for the intended purpose, raw data is never leaked, the value generated from the data is transferred to the owner). This thesis surveys the challenges unique to clinical trials, and highlights the various methods for privacy-preserving computation prior to this work. Through the overview of OPAL's solution in the space of privacy-preserving computation, we show the implementation details of how OPAL was applied to clinical trials in a project called Open Trial Chain, a platform for clinical trial data built for analytics, security, and incentivized sharing through technologies like federated learning and blockchain. With motivated examples derived from real-world reported problems in healthcare, we also demonstrate speed, accuracy, and security metrics. In the application, Open Trial Chain can drastically reduce clinical trial costs, reduce error, and increase quality of analysis diversity. Overall, this project shows promise for further extension in other health datasets for compliance in an ever-complicated move towards regulations that reflect for conscientiousness for data security, ownership, and provenance.