Multidimensional Anomaly Detection in DNA-Profile Based Personalized Medication Administration
Individual variations in drug metabolism and response are hypothesized to be due to genetic factors – therefore, determining personalized drug doses requires genetic profiling to compute the metabolic rate for a drug. These variations are due to changes such as changes in single DNA bases or single nucleotide polymorphism (SNPs), insertions or deletions of DNA bases, and the compatibilities of two alleles per gene (one maternally, and one paternally inherited).
Renaissance Rx is a biotechnology company performing high-throughput drug and genetic testing. The enormous amount of data they collect could be used to correlate genetic profiles with drug metabolism products, and, properly analyzed, could provide prescribing physicians with an unprecedented new tool to assist with personalized drug dosing and possibly avoiding unfavorable multi-drug interactions. Renaissance Rx generates ~1000 entries per day and wants to automate and detect anomalies in drug administration for their clients.
Herein, we propose a novel cluster-based anomaly detection tool, to address problems related to a very large but robust genomic dataset, in order to understand the pharmacological effect of genomic variations. We will ultimately deploy these systems to automate near-real-time anomaly detection and data dissemination to physicians. A successful implementation will aid Renaissance Rx in modernizing their personalized pharmacology pipeline, and advance our research in the area of pharmacogenomics.
Hoque, Tamjidul, "Multidimensional Anomaly Detection in DNA-Profile Based Personalized Medication Administration" (2015). Computer Science - Grants and Contracts. Paper 7.
This document is currently not available here.