Feature Story

Doctors win with a smart new addition to the team

Artificial Intelligence is upping its game in healthcare, giving medics more time to care.

Artificial intelligence (AI) is going from a simple tool to “a member of the team” in the healthcare sector, sources say. In recent months, AI has started to be used for prospective studies “on patients in the real world,” says healthcare expert Dr. James Somauroo. Before, doctors tested AI on data sets and then checked to see how the machine’s results compared to those in real life. But the AI findings were never used on patients. Now that’s changed.

“In the last four to six months, this stuff has really kicked off,” says Somauroo.

In one eye-catching example last year, a team in the U.S. got AI to review electrocardiogram (ECG) test results for signs of a future stroke or heart attack.

Not only was the neural network better at telling which patients might end up having problems within a year, but it could also spot when someone was at risk even if they seemed to have a normal ECG. Doctors don’t know how the AI was doing it.

Still, “this could completely alter the way we interpret ECGs in the future," said Dr. Brandon Fornwalt, senior author of the study, in a press note.

See also: Telehealth aids under-served communities

This is not a unique case. AI is gaining ground on human experts in many parts of healthcare, says Dr. Nigel Whittle, head of medical and healthcare at Plextek, a Cambridge, U.K.-based consultancy.

“A key element of modern medicine is the use of imaging systems for clinical diagnosis, and the main challenge for doctors is to interpret the complexity and dynamic changes of these images,” he says.

When human experts do the work, it takes time and money, and there is a risk of human error. But as AI has gotten better, “machine learning systems can extract richer information from images, with a corresponding increase in reliability and accuracy,” Whittle says.

“This will allow doctors to re-focus on their interaction with patients, which is often cited as the most valued aspect of medical intervention,” he adds.

A sign of healthcare’s changing view of AI is that one of the sector’s top journals, npj Digital Medicine, is only accepting AI-based studies that are prospective in nature, says Somauroo.

“That’s a marker in the sand,” he says.

Growing trust of AI in healthcare

The shift signals growing trust in AI within the healthcare sector, he says. This is no mean feat, since working with a machine that knows more than a human could be “a bit of an ego hit for a doctor.” But doctors cannot ignore that AI is getting results in more and more projects around the world.

In the U.K., for instance, a firm called Skin Analytics is working with University Hospitals Birmingham to pilot a scheme that lets people screen themselves for signs of skin cancer without having to see a medic, using an iPhone app. If it spots anything amiss, the AI flags the case with a doctor.

See also: Healthcare expands its reach with video conferencing

This means doctors can spend less time on cases that don’t warrant attention. But it does not mean the end of doctors as we know them. As Whittle notes, the role of clinicians will simply change. After all, AI may beat doctors when it comes to looking at data, but a machine will never match a doctor in terms of bedside manner.

“It’s a part of the team that has, arguably, no emotional intelligence whatsoever,” says Somauroo.

Hence, in the future AI may allow doctors to spend more time doing what they do best: looking after patients. And that’s good news.

“Healthcare is humans caring for humans,” Somauroo says. “Do you really want to be told you’ve got cancer on an iPhone? Absolutely not.”


The contents or opinions in this feature are independent and may not necessarily represent the views of Cisco. They are offered in an effort to encourage continuing conversations on a broad range of innovative technology subjects. We welcome your comments and engagement.

We welcome the re-use, republication, and distribution of "The Network" content. Please credit us with the following information: Used with the permission of http://thenetwork.cisco.com/.