A growing number of startups are beginning to use artificial intelligence to improve—and in some ways upend—the healthcare industry. And that might be as simple as a diagnosing a cough or as complex as curing cancer.
Cloud DX, a startup based in Toronto, for example, is using a concoction of data and machine learning to identify tuberculosis, pneumonia, upper respiratory infection and bronchitis by teaching AI to detect the difference between the diseases based on how a cough sounds. By recording 750 coughs from patients, along with 1,500 fake coughs from healthy people, the software can tell the difference between tuberculosis, pneumonia, upper respiratory infection or bronchitis.
“It’s just like Dr. House,” CEO Robert Kaul said, referencing the cranky-but-brilliant physician from the hit TV show House. “It’s all about making the algorithm smarter than the doctor’s gut.”
Cloud DX is just one of many AI startups looking to reinvent the healthcare sector—an industry that thanks to its tedious billing systems, complex treatments and expensive R&D is in need of an overhaul. According to Accenture Interactive, which analyzed data from more than 3,000 healthcare companies, the industry is in a “period of vulnerability.” That’s in large part due to weaknesses in efficiencies related to labor, transactions and innovations.
“If you look at the scale AI provides you, this industry is ripe for [disruption] because this industry is about five years behind everybody else and is very paper-intensive,” said Ritesh Patel, chief digital officer at Ogilvy’s health & wellness practice.
Ogilvy is already tracking at least 168 AI-focused healthcare startups looking to transform everything from drug discovery to sonograms. And of course, it’s not just startups that are looking to innovate. Larger healthcare companies are also experimenting more with R&D. Bayer, Johnson & Johnson, UnitedHealth and others are all investing in startups, internal R&D or pilot programs.
One company exploring the potential of AI is Merck. Last year, the drug maker partnered with Amazon to sponsor a developer competition that came up with ideas for how Amazon’s Alexa voice platform could assist diabetes patients.
AI can help with everything from revenue optimization to improving billing codes for insurance providers to decreasing fraud, according to Arielle Trzcinski, a healthcare tech analyst at Forrester. She said physicians often say they’re worried about losing their jobs to robots. However, innovations will actually free them up to spend more time—and better quality time—with patients. In fact, all of these innovations from emerging technologies like AI, voice and virtual reality will help virtual visits outpace outpatient visits by 2020.
“What’s driving a lot of it is we’re at this tipping point where patients need to be empowered as consumers and are looking for a retail-like experience,” Trzcinski said.
Earlier this year, a company called M3dicine debuted an FDA-approved technology called Stethee, which uses AI and data from heartbeats to better understand patients—a 21st century update to the stethoscope, a device that’s been about the same since it was invented in 1816. There are other AI-enabled products in homes. For example, Pillo Health reminds people when to take their pills. And just last month, AARP announced a partnership with Orbita—a platform that uses both chatbots and voice assistants to interact with remote patients—to study remote patient monitoring.
Swift Medical, a Toronto-based startup, is working to change the way wounds are analyzed and treated. With a smartphone app, the company uses AI and machine-based learning to collect images and measure variables. The software can see the length and depth of a wound, which then flows into a cloud-based backend to help health providers move forward with a plan of action. That’s especially helpful—considering there are 280 variables used to understand whether a wound is healing or not, according to Swift Medical CEO Carlo Perez.
“We like to call it an automation of wound care from capture to claim,” he said.
However, it’s not as easy to go to market with a healthcare app as it might have been for Candy Crush. Regulations protecting people from products are sometimes a hurdle for companies to cross. In the case of Cloud DX, receiving FDA approval for algorithms takes between six and 12 months. And, unlike the average iPhone app, each and every change requires its own clearance as well.
“When you think about that cycle, it’s no wonder there are very very few medical device startups that have scale,” Kaul said. ”On the other hand, who ever gets it first and gets it right kind of has the business to themselves.”