John Halamka: Telehealth, Apps & AI Make Progress, But Will Everyone Benefit?

Xconomy Boston — 

John Halamka thinks the digital health industry is still “emerging.” But it has come a long way and is starting to deliver after years of hype.

Halamka, a Boston-based physician and healthcare technology expert, says that’s thanks to several coalescing factors: improved technology, more favorable financial incentives for using digital products in healthcare, and growing demand from patients accustomed to tech-enabled convenience in other areas of their lives.

“In 2019, the tech is better, but also the alignment of incentives is better,” says Halamka, who leads innovation at Beth Israel Lahey Health system and is a Harvard Medical School professor.

Over his 30-year career in medicine, Halamka has had a front row seat to advances in healthcare technology, and at times has helped drive them. When he arrived at Beth Israel two decades ago, he pushed it to move its homegrown electronic medical records system to the web—years before “the cloud” entered the tech lexicon. Watertown, MA-based healthtech company Athenahealth acquired Beth Israel’s system in 2015.

Last year, Halamka was named executive director of Beth Israel’s new Health Technology Exploration Center, where a team of clinicians, engineers, and project managers experiment with artificial intelligence software, blockchain systems, and other tools that might have an impact on healthcare.

He says he’s also traveling a cumulative 400,000 miles each year, visiting countries such as Japan and Israel to scout technologies and strategies that might improve healthcare in the US.

All this to say, Halamka has his finger on the pulse of healthcare technology development and implementation worldwide. Xconomy spoke with him recently to take stock of digital health’s progress and map out where it might be headed next. (You can hear more from Halamka on Oct. 22 at “X·CON 2019: Digital Health Gets Real,” an Xconomy conference being held at deCordova Sculpture Park and Museum near Boston.)

Here are four more takeaways from our conversation, spanning the potential of wearables, apps, and AI in healthcare—and their possible pitfalls:

1. Nobody really knows exactly what to do with the tons of health data being collected.

Millions of people wear Apple Watches, Fitbits, and other internet-connected devices that can track their heart rate and other biometrics with health implications. Now what?

“We’re generating all these novel sources of data, but nobody knows quite what to do with it, where to store it, what to trust … and what are the legal and ethical complications,” Halamka says.

For example, let’s say a doctor’s office offers a service where it will store a patient’s wearable device data in his or her electronic health record. But what if the doctor never looks at the wearable data? What if the doctor doesn’t take immediate action if the data seem to indicate an emergency? What’s the right threshold for acting upon the data? “No one has really defined these issues,” Halamka says. “Who is accountable for reviewing and taking action on these novel sources of data?”

It will be important for the medical industry and other stakeholders to develop a set of “community standards” for handling data from wearables, machine learning algorithms, and other digital technologies, Halamka argues.

2. AI won’t replace your doctor. But it is making strides as an assistant in diagnosis and healthcare operations.

The blustery hype around AI’s potential in healthcare has died down somewhat, but the technology is making progress. As an example, Halamka points to Alphabet-owned Google’s (NASDAQ: GOOGL) research demonstrating the capabilities of machine learning algorithms to help diagnose conditions such as lung cancer.

“It’s not that algorithms are smarter [than humans], it’s just that they’re given much more data than a human would have the opportunity to digest,” Halamka says. “They’re very good at helping steer us in a general diagnostic direction.”

Another area of AI advances in healthcare is a boring, but often useful, one: clinical workflow improvements, Halamka says. By that, he means using advanced data analytics to try and predict, say, the chances that a patient will be discharged from the hospital today, or the likelihood someone will show up for their doctor appointment.

“With those kind of workflow refinements, machine learning has worked pretty well,” Halamka says.

So, will AI replace your doctor someday?

“Mmm, probably not,” Halamka says. “Machine learning is very good, but empathy and respect and active listening—that would not be the first use case I would pursue.”

One big opportunity he sees is developing AI technologies to … Next Page »

Single PageCurrently on Page: 1 2