Drug repurposing : design, emulation and analysis of synthetic in-silico clinical trials using electronic health records and modern data analytics

Cancer has been a worldwide health issue, and its burden is considered to increase in the future. For most cancer disorders, the success with current therapies has been limited. Even after huge investments in drug development, the need for therapeutic advances remains high. As effective anti-cancer drugs are in high demands, drug repurposing, using existing drugs for other diseases has sparked a growing interest. Drug repurposing presents a striking opportunity and potentially significant cost-saving in the future treatment of cancer. The cost and complexity of conducting randomized clinical trials (RCT), the growth of electronic health record (EHR) sources, and the thriving technological advances in modern data analytics create an unparalleled opportunity to develop a systematic approach for drug repurposing,using EHR data and sophisticated analytical methods. In this thesis, by leveraging enriched high dimensional EHR data with diagnosis, drug prescription and lab test information, we aim to develop a systematic approach to emulate clinical trials regarding various drugs and diseases based on modern data analytics. Specifically, we take a data-driven approach to repurpose anti-diabetic drugs for several types of cancer incidence and mortality risks among the aging population, through the lenses of optimization, statistics, and machine learning. We start by introducing background knowledge for this study including cancer, drug repurposing, anti-diabetic drugs and clinical trials in Chapter 1. In Chapter 2, we describe the UK primary care database Clinical Practice Research Datalink (CPRD) along with its data structure for data preprocessing. Methods and mechanisms for missing data in clinical studies are also discussed as they will influence model robustness, statistical significance and directional results. In Chapter 3, we discuss alternative frameworks for survival analysis and causal inference with emphasis on modelling the behavior of how physicians prescribe drugs, using propensity scores. Several Cox regression based semi-parametric methods are also reviewed for survival analysis. Chapter 4 offers baseline characteristics for a comprehensive insilico randomized controlled trial with a total of 640 model specifications. Chapter 5 presents numerical risk ratio results for 10 sub-studies and discussions of covariate balance evaluation and sensitivity analyses among 64 schemes within each sub-study. Through this work, we have made preliminary contributions to repurposing anti-diabetic drugs for cancer incidence and mortality risks. More importantly, we have offered a systematic approach that has the potential to be used to repurpose drugs for other diseases that are of interest. This use of modern data analytics offers tremendous potential to meet healthcare challenges in this era of rapid technological change.

Name thesis:degree_name

Master

Department dc:contributor.department

Massachusetts Institute of Technology. Department of Mechanical Engineering

Grantor dc:publisher

Massachusetts Institute of Technology

Year dc:date.issued

2019

Author dc:creator

• Xu, Shenbo, author.

Advisor dc:contributor.advisor

• Roy Welsch and Stan Finkelstein.

• Mechanical Engineering.

Statement dc:rights

• MIT theses may be protected by copyright. Please reuse MIT thesis content according to the MIT Libraries Permissions Policy, which is available through the URL provided.

Licence dc:rights.uri

• http://dspace.mit.edu/handle/1721.1/7582

Language dc:language.iso

eng

Handle dc:identifier.uri

https://hdl.handle.net/1721.1/150468

OAI identifier oai:identifier

oai:dspace.mit.edu:1721.1/150468