Paging Dr. Robot: How AI Is Saving Lives and Disrupting Healthcare

Here, in an article originally published on BRINK, CEO and Founder of Skymind, Chris Nicholson, offers his expert perspective on the future of artificial intelligence in healthcare.

“Robot doctors” are working alongside their human counterparts, making real-time predictions in life and death situations, and saving lives in the process by helping doctors identify better treatment options.

These “robots” are actually artificial intelligence systems designed to comb through the vast amount of data being generated through electronic health records and electronic medical records at a pace impossible for human doctors to match. Advanced clinical analytics are making predictions about the outcomes of various healthcare treatments and are helping to train doctors to make accurate predictions faster.

Today we have more data than we know what to do with. The healthcare industry—now focused on artificial intelligence instead of business intelligence—must focus on gathering the right data and aggregating this data to make it transparent and actionable.

Better Care, Lower Costs

Many forces have combined to give healthcare organizations an opportunity to provide more affordable, higher-quality care. The powerful algorithms that tap into big data and real-time data streaming engines, for instance, are already helping empower people to make better rapid-fire decisions about different facets of care delivery.

As it evolves, artificial intelligence will reduce human error and help those in healthcare make critical predictions and decisions about high-risk or high-cost patients, adverse health events, hospital readmissions and more.

Artificial Intelligence Equals Math — With a Kick

Perception and interpretation lend meaning to raw sensory data, such as the images your eyes see at this very moment. Something in our brains puts a name on what we see. This is exactly what algorithms and artificial intelligence can do, too, with superhuman accuracy. Machine perception is also applicable to data such as sound, voice, time series and text.

Most computer programs execute precise commands. They do what they are told. But certain algorithms can be taught to perform better over time on well-defined tasks. Those algorithms learn and can rewrite themselves in response to the data they’re exposed to. We are fascinated with artificial intelligence because of its potential to be smarter than its creator, learning from its mistakes along the way.

AI, machine learning and deep learning are used a lot, and sometimes interchangeably, but they have distinct meanings, because they are subsets of one another. All machine learning is artificial intelligence, but not all artificial intelligence is machine learning. And all deep learning is machine learning, but the reverse is not true. Artificial intelligence can be static; i.e., it doesn’t necessarily change in response to data. But machine learning can beat you at checkers because it learns from its “mistakes.” Deep learning can achieve superhuman-level accuracy. And changes in the accuracy of our predictions will affect the life and death of companies and the life and death of patients.

Feed the Beast

Big data is getting bigger. The quantity and types of data being created are exploding. This allows deep learning—which requires tons of data to make accurate predictions—to produce tremendous value for healthcare organizations. But we need to gather the right data and process it correctly. The data should be aligned with the outcomes we want to predict. We need to correlate and normalize all possible information about a single patient and feed it into a single artificial intelligence system to make an accurate prediction about that patient’s health.

Getting good at these kinds of predictions saves lives. Accurate predictions about whether or not people will die, for example, will transform the business of intensive care units—achieving extraordinary results with ordinary effort.

For example, researchers in Boston produced CT scans, looking for signs of cancer. Algorithms not only got good at predicting cancer, but they also generated innovative developments in science by helping medical practitioners recognize new features and symptoms to look for in patients.

When doctors are able to align all available patient data with the outcomes they need to predict—and then be given accurate predictions about treatments and results—the true power of artificial intelligence can be realized.

Click here to watch the full main stage session during this year’s Oliver Wyman Health Innovation Summit.