SFU computing science professor Martin Ester is mining the full clinical and genomic adverse drug reaction (ADR) dataset in an effort to predict and prevent severe ADRs in children undergoing cancer treatment.

Using data science to predict and prevent adverse drug reactions in children with cancer

May 07, 2018
Print

Cancer treatments can often lead to adverse drug reactions (ADRs) which not only affect the patient’s recovery but can also lead to permanent disabilities and death. For children undergoing cancer treatment, the risk of severe ADRs is even greater.

“ADRs in children are three-fold more likely to be life-threatening,” says Martin Ester, a professor in SFU’s School of Computing Science.

“A staggering 75% of childhood cancer patients develop chronic health conditions and 42% develop disabling or life-threatening ADRs from cancer treatment.”

Ester is a part of a research team that was recently awarded $9.9 million through Genome Canada’s 2017 Large-Scale Applied Research Project Competition: Genomics and Precision Health. The team is led by Drs. Bruce Carleton and Colin Ross from B.C. Children’s Hospital who are studying how a patient’s response to medication is influenced by their genes, a field known as pharmacogenomics.

Ester, who was named the world’s most influential data mining scholar in 2016, is leading a sub-project to mine the full clinical and genomic ADR dataset using a machine learning approach.

“We have collected over 6,125 DNA samples and have corresponding reports of medication use and ADR outcomes recorded in a clinical database,” says Ester. “This is the foundation that will help us identify new connections between genetic variations and ADRs and develop algorithms to predict a child’s likelihood of suffering an ADR during their cancer treatment.”

Physicians could use the technology to asses a child’s susceptibility to ADRs ahead of time and provide personalized treatments that would prevent serious ADRs and save lives.

Establishing accessible pharmacogenomic screening in Canada

While the current dataset of 6,125 DNA samples are ample enough for the development phase of the project, moving forward the research team will require a larger dataset that is more representative of the population and able to ensure accurate ADR predictions. However, pharmacogenomic testing poses a challenge as it first requires both physician and patient education to support informed-decision making.

To address this need, the team is also developing tools and resources to assist with knowledge translation and application of their findings into clinical practice. And as they expand pharmacogenomic testing across Canada to support their research, they hope their efforts will help establish accessible pharmacogenomic screening in the country.

“We want to provide this much-needed access to pharmacogenomic testing in pediatric oncology in Canada,” says Ester.

“This data will build the foundation for personalised medicine and for improved individual and population health.”