MIT professor's quick primer on two types of machine learning for healthcare
There are two main approaches to machine learning – supervised and unsupervised – and each has specific applications in the context of healthcare.
And even though their impact has not yet sent shockwaves through the industry, the potential of each is enormous, according to John Guttag, head of the Data Driven Inference Group at MIT's Computer Science and Artificial Intelligence Laboratory.
At its basic level, machine learning involves looking at data, and from that data finding information that is not readily visible. Example: Applying machine learning to data about patients infected with Zika or another virus and using what we can learn about what happens to those people to inform care decisions regarding the best ways to treat people who get infected in the future.
"Typically we use machine learning to build inference tools, where we find patterns in existing data that allow us – when presented with new data – to infer something interesting about that data," said Guttag. “Machine learning is driven entirely by the data, rather than by, say, human intuition."
Here’s a look at the two main types of machine learning and why they matter to healthcare.
Supervised machine learning
"In supervised machine learning, we're given the data and some outcome associated with the data," Guttag explained. "We're given all the people who have Zika infections and then we know which of the women have children with birth defects and which don't. And maybe from that we could build a model saying that if the woman is pregnant and has Zika, what's the probability that her baby has a birth defect. And it might be different for 30 year old women than for 40 year old women. Who knows what the factors would be. But there we have a label – all sorts of details about the woman, and was the baby healthy or not. So that would be supervised learning: We have a label about the outcome of interest."
Unsupervised machine learning
Unsupervised learning, on the other hand "means we wouldn't have a label," he said. "We just get data, and from that data we try to infer some hidden structure in the data. So for example you get a bunch of healthcare data and you find patients who look 'similar.' Typically the nice thing about unsupervised learning is you find things you weren't even looking for. It's also useful for when, for one reason or another, the data is impossible to label."
The case for using emerging tech today
Guttag added that machine learning is among the fastest growing parts of computer science right now in the world. As healthcare entities continually ramp up their analytics and big data efforts and gird for precision medicine and population health, machine learning as well as artificial intelligence and cognitive computing are poised to become even more valuable.
While vendors such as IBM Watson, Google, Microsoft, and other tech giants are bringing new technologies to market, most of the progress made in machine learning is happening in financial services, retail and other industries, and has been for about a decade.
Healthcare, true to its reputation for slowly embracing new technologies, is a bit late to the party.
One of the challenges unique to healthcare is the long gap between when new knowledge is obtained and when clinicians and doctors can put it to use treating patients, which is among the reasons Guttag urged major healthcare providers to more aggressively integrate today’s machine learning tools into their workflows now.
“People should be using today's technology to do things today," Guttag said. "Machine learning is a huge deal. And we're going to see some pretty dramatic changes over the next few years.”
Related stories ahead of Big Data & Analytics Forum in Boston, Oct. 24-25.
⇒ Tips for reading Big Data results correctly
⇒ Small hospital makes minor investment in analytics and reaps big rewards
⇒ MIT professor's quick primer on two types of machine learning for healthcare
⇒ Must-haves for machine learning to thrive in healthcare