May 1, 2014

Big Data Helps Personalize Cancer Treatments

George Leopold

There’s a new tool in the fight against cancer and other diseases with a DNA link: big data.

The Cancer Institute of New South Wales and the Children’s Hospital at Westmead in Australia have awarded a “Big Data, Big Impact Grant” to researcher David Skillicorn of the School of Computing at Queen’s University School of Canada. His research focus will be applying big data techniques to help “personalize cancer treatment for children.”

The grant will support research by Skillicorn and ten other investigators on big data applications related to childhood cancer patients suffering primarily from leukemia. The project is titled, “Generating Actionable Knowledge from Complex Genomic Data for Personalized Clinical Decisions.”

Current technology, called “high-throughput devices,” collects tens of thousands of marker values for each patient. Patients are then clustered and their eventual treatment is based on their cluster. 

The big data research could result in a redefinition of these clusters.

“Patients don’t form clusters,” Skillicorn said in a statement released by Queen’s University. “The disease almost always looks different from one patient to another. We believe there must be some bottleneck that causes the wide variety of patient configurations to appear as a much smaller set of disease categories.”

Big data techniques could then be applied to come up with the appropriate treatments for childhood leukemia patients.

As efforts like the Human Genome Project generate vast amounts of data, they are providing deeper insights into the genetic underpinnings of diseases like leukemia along with progress toward the development of a new generation of therapies. One result has been the rise of “translational bioinformatics” as an emerging discipline that builds on bioinformatics and big data in the study of complex diseases.

While techniques like DNA sequencing have been used to treat leukemia patients, the application of big data can sometimes slow progress because doctors must correlate data from full genome sequencing with mountains of data generated in studies and medical records. The challenge is complicated by the fact that medical information is doubling every five years.

Big data companies like IBM Research are also working on applications that could be used to personalize cancer treatment. The company’s Computational Biology Center is working on targeted cancer therapies based on individual patients and the genetics of their form of cancer.

IBM researchers predict that within several years cognitive systems that sift through terabytes of medical data will be able to help doctors apply big data techniques to unlock DNA sequences that could be used to pinpoint cancer therapies for individual patients.

The hope is that cognitive systems that continuously learn more about cancer and cancer patients, eliminating the clusters that are currently used to determine treatment. This approach also could be used for other diseases with a DNA link such as heart disease or stroke, IBM said.