UWM researchers use big data to solve big problems

By Laura Otto, Angela McManaman and Kathy Quirk
UWM Report
June 21, 2018

Big data. On the one hand, it’s information overload. On the other, it’s gold in the quest for faster, smarter tools that can be applied to automation – and just about everything else.

That deluge of data is ripe for enabling artificial intelligence, machine learning and prediction analytics, collectively known as data science, which can find patterns that would have otherwise been missed by human beings.

Data science uses computational techniques that blend historical knowledge with uncertain elements to provide sophisticated inferences.

Data scientists build complex models that can make predictions in dynamic conditions, recognize speech and faces, transform images, improve drug development, and support investment and business decision-making. UWM data scientists work in disciplines across many of the university’s schools and colleges.

And now, UWM is poised to make a bigger commitment to big data with the establishment of the Northwestern Mutual Data Science Institute. Northwestern Mutual and its foundation will contribute $12.5 million, while UWM and Marquette University will each contribute in kind and fundraising of $11.25 million over the next five years. The investment will support endowed professors, data science faculty positions, research projects and expanded student programming, which will begin this fall.

Here’s a look at some of the UWM scientists already using data science in their work.

Chiang-Ching (Spencer) Huang

Chiang-Ching (Spencer) Huang is using big data to search for a better way to detect and treat cancer.

The ultimate goal of the work is to bring liquid biopsies – tests of bodily fluids, such as blood – into clinical practice. These tests would be less intrusive and provide more information than current biopsies.

Scientists already know fragments of tumor DNA are released into the blood. “We want to know how accurate tests are in detecting cancer cells in the blood,” said Huang, an associate professor biostatistics in the Joseph J. Zilber School Of Public Health.

Using DNA extracted from the blood of healthy patients and those with cancer, Huang is using data mining to sort through “humongous” amounts of genomic data to identify robust biomarkers linked to certain types of cancerous cells.

Being able to detect tumor cells earlier gives doctors an indicator of potential cancers before symptoms appear, and can also help enhance the effectiveness of treatment.

“We measure glucose and cholesterol in routine screenings,” Huang said. “When we are able to find cancers in a routine check in the very early stages, they are much easier to treat.”

Paul Auer

Paul Auer doesn’t have a scalpel or a stethoscope, but his work is having an impact on the treatment of diseases.

Auer, an associate professor of biostatistics in UWM’s Joseph J. Zilber School of Public Health, uses mathematics and computers to study millions of genes to try to isolate variations that are linked to heart disease, cancer, sickle cell anemia and other diseases.

This type of big data statistical detective work is becoming an increasingly important part of health care as researchers comb through genetic information and large-scale, long-term compilations of medical records to establish links between genes and disease.

Most recently, he has been part of an international team publishing findings on genetic variations related to a certain type of estrogen-negative breast cancers. He’s also part of a study, recently published in Nature, on protein-altering variants associated with body mass index. Those findings could eventually help lead to better therapies for obesity.

Read the full article here.