Google has created an artificial intelligence (AI) system to predict a range of potential patient outcomes, including when a patient might die, Mark Bergen reports for Bloomberg.
How it works
Google's technology uses neural networks, a form of AI that can use large amounts of data to learn and improve itself. Google's technology can dig through the voluminous data that providers create about patients—including notes saved in PDFs or in old patient charts—and make predictions faster and more accurately than current methods, Bergen reports.
According to Nigam Shah, an associate professor at Stanford University and a co-author on the research, as much as 80% of the time spent creating predictive models currently goes to what Shah called the "scut work" of making data presentable. Google's technology, however, is able to avoid this entirely. "You can throw in the kitchen sink and not have to worry about it," Shah said.
According to Vik Bajaj—a former executive at Verily, the health care arm of Alphabet, which is Google's parent company—Google's new technology can "leapfrog" other software because it can learn to look through data on its own.
In one instance, research shows that Google's technology predicted that a patient with late-stage breast cancer had almost a 20% chance of dying, while prior technology estimated only a 9% chance of death. The woman died "in a matter of days," Bergen reports.
How Google plans to move the tool into clinics
Google now is look to bring the technology into a clinical setting, according to Jeff Dean, Google's AI chief and leader of a research unit often referred to as Medical Brain. Dean said Medical Brain is working on a number of different AI tools that can predict and diagnose symptoms and diseases with accuracy.
For instance, Bergen reports that Google is working on AI systems for cardiology, ophthalmology, and radiology. The company also has started working on an AI system for dermatology, with an app that can spot malignant skin lesions. Google could license these systems to clinics or sell them as a type of "diagnostics" service, Bergen reports.
Exciting potential, but some concerns
Bergen reports that Google is excited about the potential of this new technology, especially since it gives the company a way to move further into the health care market—something it has been attempting to do for a long time.
Dean hopes the new AI systems ultimately will guide doctors toward specific medications and diagnoses.
However, some experts are concerned about the amount of data Google will access through this new technology. Andrew Burt, the chief privacy officer for Immuta, said Google and similar companies "are going to have a unique, almost monopolistic, ability to capitalize on all the data we generate."
For its part, Google says the patient data it utilizes is anonymous, secure, and being used with patients' permission, Bergen reports. But Samuel Volchenboum, a pediatric oncologist, said Google likely will have difficulty maintaining that rigor if its AI systems expand to smaller hospitals.
Despite those concerns, Volchenboum believes that the AI systems have the potential to save both lives and money, using data such as the local weather and traffic to determine potential patient outcomes (Bergen, Bloomberg, 6/18).
Big data: It seems impossible to avoid, but is it right for you?
In this primer, we examine what big data is, the problems big data technologies attempt to solve, and the vendor landscape. We also provide guidance on when and if it makes sense to deploy big data approaches in health care settings.