Preliminary EHR Data Analysis Toward Developing Precision Oncology for ColoRectal Cancer

Research Paper
Authors:Veeraraghavan, Malathi, EN-Elec/Computer Engr DeptUniversity of Virginia ORCID icon, MatthewUniversity of Virginia Badrinarayanan, Vijay Lingesh, EN-Comp Science DeptUniversity of Virginia

There is well established evidence of inter- and intra-patient variability in pharmacokinetics (PK) when Fluorouracil (5-FU), a key component of chemotherapy treatments for ColoRectal Cancer (CRC), is dosed by body surface area. Toward developing precision oncology methods that leverage Electronic Health Records (EHR) to customize dosing for patients, we started a data-analytics project for CRC patients. This document describes successes and failures encountered in this preliminary effort. We successfully procured a dataset consisting of five files: patient information, medications/doses, labs, vitals and Cancer Registry data. However, the dataset was incomplete and we could only extract sufficient information for 94 patients out of a starting set of 1460 patients. But we made important advances in defining single scores for complex multi-drug chemotherapy regimens, handled the problem of irregular time series, and were able to visualize the efficacy and toxicity of treatment for these 94 patients. We also made preliminary advances in creating subsets of patients, an initial step toward developing predictive models.

Precision Oncology, Electronic Health Records, Colorectal cancer
University of Virginia
Published Date:
December 31, 2019