Wearables long way from predictive analytics, says Intel chief
September 8, 2015
Although wearables are helping health providers, organisations and individuals monitor medical activity, there is still a long way to go before they can provide serious predictive analytics, according to Christopher Gough, a lead architect at Intel’s health and live sciences group.
In a blog post, he said that wearables had become more than a passing trend and were truly changing the way people and organisations thought about managing health.
“I hear from many companies and customers who want to understand how the wearables market is impacting patient care as well as some of the changes taking place with providers, insurers and employers,” he said.
For providers, he gave the example of a pilot that the Mayo Clinic did with Fitbit to track patients recovering from cardiac surgery. They were able to predict which of those patients would be discharged sooner than others based on their activity in the hospital.
“You can easily see how this use case could be extended outside of the hospital, where you might be able to use wearables to more accurately predict which patients are at the highest risk for hospital readmission,” he said. “This of course is a key quality metric that hospitals are incentivised to reduce.”
On the payer side, organisations are using wearable devices to influence the behaviour of their members, encourage a healthier lifestyle, and delay the onset of conditions such as obesity and diabetes. Cigna has a programme for their own employees where it identifies individuals who may be at risk for diabetes.
“They created a wearables programme that encouraged increased activity in those individuals’ daily lives, and it’s making a difference,” said Gough.
Gartner says that more than 2000 corporate wellness programmes have integrated wearables to track employees’ physical activity and incentivise them, sometimes financially, to have a healthier lifestyle. BP rolled out a programme with 14,000 employees. Those who were able to achieve one million steps – equivalent to roughly 800km for an average-size person – over the course of a year received a health plan premium reduction the following year.
But he said nobody had been able to aggregate enough wearable data for some serious predictive analytics.
“I think that’s down the road,” he said, “certainly before it becomes mainstream. This will entail significant data integration and big data analytics. We’re looking to pull in multi-structured data from multiple distributed entities and repositories – data from electronic health records, health insurance claims, in some cases socioeconomic data, and all the new sensor data from wearables.
“If we can pull the continuous stream of patient-generated data into a repository, and overlay more traditional payer and provider data, I suspect the accuracy of predictive models will be significantly improved. We’ll be much better able to identify high-risk patients that will benefit most from additional outreach by a provider organisation.”