• 2019 Family Medicine Experience

    AI Moving Health Care to ‘Cusp of a Major Revolution’

    September 30, 2019 05:41 pm David Mitchell Philadelphia –Punit Soni spent more than seven years with Google, rising to lead product manager before serving as vice president of product management for the company's Motorola division. He eventually moved on to Indian e-commerce giant Flipkart, where he spent a little more than a year as chief product officer.

    But Soni wanted to build something of his own, so in 2016 he moved back to the United States and began pondering what he could build with emerging technologies. The result is Suki, a digital assistant for physicians that combines artificial intelligence and voice-enabled technology.

    "When I was trying to decide what to do with these platforms, I was shadowing a family physician in a hospital," Soni, Suki's founder and chief executive officer, said Sept. 26 during a main stage panel discussion at the Family Medicine Experience. "I was amazed to see all the clicking of boxes and how feverishly they were putting everything together. We took a sophisticated group of professionals and turned them into data clerks. This is what technology has done to you."

    Soni has a better idea. He gave a demonstration of how a voice-enabled assistant can pull a patient's history, take notes, order prescriptions and save the updated file to an EHR system.

    Story Highlights

    Panelists discussed artificial intelligence in health care during a main stage event at the Family Medicine Experience.

    Punit Soni described Suki, a digital assistant for physicians from his company, which is a partner in the AAFP's project to find solutions to EHR-related problems.

    Lily Peng, M.D., Ph.D., product manager for Google AI Healthcare, discussed how physicians can work with artificial intelligence.

    As administrative burden is one of the top concerns of family physicians, according to the AAFP's annual Member Satisfaction Survey, the Academy launched a 42-month special project to address issues related to EHRs -- and to work toward solutions -- last year. Suki is one of the Academy's partners in that project, and the company also participated in the Office of the Future exhibit at FMX.

    Soni, however, emphasized that Suki is not a futuristic prototype. He also acknowledged that there are several companies working toward these types of solutions for physicians.

    "These are the kinds of technology that can be used so that you can focus on what you love to do," he said. "This exists. This is an actual product."

    He said digital assistants should be "invisible" so doctors can focus on patient care.

    "I hope this works better than the last round of tech that showed up in the 1990s," he said, referring to EHRs. "What if I told you AI will fundamentally change everything we do? We are on the cusp of a major revolution in tech."

    So how can companies ensure that this next iteration of health IT is different? Soni said a key is to develop products in a physician-led manner. It is important that physicians and technologists work together, focused on delivering a product with its end users in mind.

    "A key aspect of a product like Suki is having a keen understanding of doctor workflows," he said. "How does the doctor practice medicine?"

    Similarly, Lily Peng, M.D., Ph.D., product manager for Google AI Healthcare, said her team combines engineers, scientists and physicians.

    Peng said her team developed an AI tool to detect diabetic retinopathy, and it reduced false positives and false negatives 11% and 5%, respectively, compared to eye specialists.
    That doesn't mean the technology will replace physicians, she said. For example, Google also developed an AI tool to screen for breast cancer. The tool found 95% of lesions with eight false positives, compared to physicians who spotted 73% of lesions with zero false positives. Using a combination of physicians and the tool was more effective than using either one alone, she said.

    While Soni said Suki is available now, Peng said the screening tools Google is working on are years away from widespread clinical use.

    "When I see my patients and I don't think about it being a tool -- I just think, 'I got my job done' -- that's how we'll know AI has made it," she said.