How big data can improve healthcare

With the world swimming in data—from endless Twitter streams and Facebook updates, to genetic tests and medical records—it’s up to computer scientists like UBC Computer Science Professor Raymond Ng to help make sense of it all. Ng, Chief Informatics Officer for the PROOF (Prevention of Organ Failure) Centre of Excellence, discusses how big data is revolutionizing medicine, and why privacy is of utmost concern.

How can big data improve healthcare?

We have a lot of new bio-technologies that allow us to thoroughly look at a patient at the molecular level—such as DNA aberrations, changes in gene expression and protein expression. Just one sample from a patient at one time point can easily generate hundreds of thousands of readings. If you have 1,000 patients and 10 time points, in just a single data set you have over a billion numbers to analyze. That’s why we need computational methods to analyze the individual deep data.

This all fits into personalized medicine; to make a diagnosis or prediction of whether an individual is going to have a certain condition, you need very deep data on that particular individual.

Is this being used in medicine today?

It’s beginning to be, particularly in cancer clinics where they try to predict, for example, whether someone is going to respond to a certain chemotherapy. But we are still barely seeing the tip of the iceberg. Personalized medicine will have a lot more clinical applications in the next 10 to 20 years.

One of the tests that we are developing at PROOF is to take a tube of blood from a COPD [Chronic Obstructive Pulmonary Disease] patient, at a time when the patient may have some early symptoms, to try to predict whether this person is going to have an exacerbation of their disease, or a lung attack, within 60 days.

If a test like this is successful, then the clinician may start giving the patient more powerful medication, such as steroids, before the lung attach occurs. This would, hopefully, keep the patient from being hospitalized. As well as helping patients, such a blood test would save the healthcare system a lot of money.

In typical health research, patient privacy is of utmost importance. All of this work undergoes very rigorous ethics approval both locally and at the federal level.

Beyond the context of specialized medicine, there are very powerful techniques that have been developed, typically under the banner of data mining, machine learning, or computational statistics. I could see that when access to this big data gets into the hands of the wrong people, it also could do a lot of damage. We have the technologies to collect and analyze big data, and I think we need to make sure that the governance structure is there to minimize abuses.

"If you have 1,000 patients and 10 time points, in just a single data set you have over a billion numbers to analyze. That’s why we need computational methods to analyze the individual deep data."

Alex Walls
Media Relations Specialist, UBC Media Relations
alex.walls@ubc.ca