A group of top developers in medical AI discussed how the technology will transform medicine
Medical professionals will be able to use artificial intelligence to diagnose patients and decipher the best method of treatment within the next 10 years, according to the co-founder of a company that provided health analytics tools to hospitals.
Allen Kamer, who helped found Humedica, a US-based company that later reportedly sold for “hundreds of millions,” according to the Boston Business Journal, claimed that within the next decade, medical AI will be able to provide what he described as “personalized medicine.”
Speaking at the eighth annual Geektime Conference in Tel Aviv, Kamer told the audience that, “In the long term, I imagine a scenario where when a patient goes to their physician with their symptoms, there’s a lot of other external data that comes with that patient that’s informing the physician. Everything from behavioral information, [to] how the patient will respond to particular types of treatment.”
“I don’t just mean pharmaceutical treatment, but I mean a digital intervention. I mean the type of clinician that would interact with that patient, and really, a whole profile of what’s going to work best for this type of patient. AI comes to personalized medicine. I think it’s a decade out but I think the toolsets … are really starting to permeate the market.”
Kamer said that within the next five years, this technology will reach specific diseases, and broaden from there. “Shorter term than that you’ll see … examples of specific diseases where patients will really benefit from different types of interactions, rather than triage and acute care. Having more time and more in-depth information will enable different care pathways earlier.”
Kamer appeared at the event alongside Arturo Weschler, the chief medical officer of Medial Research, a private research institute in Kfar Malal that uses machine learning to make predictions on which patients are likely to contract certain diseases. Weschler said that it will be patients, and not practitioners, who will embrace the ubiquity of AI in medicine.
Weschler believes that AI “will empower patients” by allowing them to monitor their diseases. “Even if the physician is resistant to the change, the change will be brought by the patient,” he said.
However, medical AI faces challenges. Ronen Gadot, CEO of ElMinda, a Herzliya company that created a tool that can visualize serious brain diseases such as Parkinson’s, said a major challenge is that data is often painstakingly difficult to collect. “Some data are available out there in electronic medical records. Sometimes it’s written by hand.” Gadot revealed to the audience that, “It took us 10 years to collect the first 30,000 datasets.”
Allen said another difficulty is getting physicians to accept what the data says. “Physicians have to feel confident that [the data] is telling them something that they can stand behind. There’s liability, there’s the possibility of harming a patient. There are lots of steps along the way that create adoption problems that have caused this to perhaps lack other industries like fintech or cyber.”
While Weschler described medical AI as “like selling Uber to taxi drivers,” Eyal Gura, the co-founder of Zebra Medical Vision, which will soon announce an algorithm that can detect risk for compression fractures, said the growth of driverless cars has actually helped medical AI get the attention of investors. “When we go to conferences, people think medical AI is around the corner.”
“You need investors and employees to believe in it,” he says.
Still, fears persist in this field that are not present in other industries. Allen says that, “You can foresee things going down the wrong path” if companies don’t carefully consider what these new systems say. “The attention to detail, the time spent really segmenting the data and breaking it down to its innermost parts is so critical. And those organizations that don’t do that, that would be a concern to me. You need to make sure you have the right quality control mechanisms in place to ensure you have appropriate healthcare.”
Gadot raised a more obvious concern: privacy. “Some of our colleagues found with using machine learning and looking at brainwaves, they can find a signature [which shows] if a person is an alcoholic or not. If you give away your brain waves, do you really want some people to know you have an alcohol problem? Your insurer probably wants to know that.”
While the same kind of exposure could come from a basic paper medical file being leaked, the issue of data security – or lack thereof in many cases – could have many worried about the continued privacy of such personal details.