Rick Newell, MD MPH is CEO of Inflect Health, Chief Transformation Officer at Vituity, and passionate about driving change in healthcare.
getty
As both an executive for a healthcare innovation and investment hub and a practicing emergency physician, I often see firsthand how machine learning and artificial intelligence (AI) technologies could bring improvements in patient care while also benefiting physicians, other healthcare providers and payers. Although there are challenges, I believe the potential upside is worth investing in.
Recently in the emergency department, I cared for a 50-year-old male patient with nonspecific fatigue. I found that he was anemic and had slightly elevated creatinine and calcium. I recognized the pattern as consistent with multiple myeloma, a cancer of the bone marrow. Sometimes, though, patterns like this can go unrecognized. Software that correlates symptoms, physical exam findings, laboratory testing and imaging into possible diagnoses could help physicians keep from missing illnesses.
In the history of mankind, I don’t think there has ever been a more powerful way to solve problems with novel insights than through the use of AI. Its ability to solve fantastically complex mathematical problems and discover the subtle and nuanced relationships between data has incredible potential to improve healthcare.
Detecting And Identifying Disease
AI can correlate the separate experiences of thousands of clinicians to identify patterns of diseases for which knowledge is only emerging. For example, in the heat of the pandemic, when doctors began widely using telehealth methods, I cared for a child who’d been diagnosed with Covid-19 and recognized additional symptoms similar to Kawasaki disease. Only later was this new clinical syndrome recognized as multisystem inflammatory syndrome in children (MIS-C), which can present like Kawasaki disease in Covid-19 patients.
AI could have aggregated these “hmm, this is odd” observations of widespread telehealth doctors to the earlier identification of a disease that we didn’t yet know existed. It could also identify population-wide diseases tied to genomics and environments, patterns which can take years for individual physicians to recognize by comparing experiences across each of their individual practices.
Machine learning has already proven effective at disease identification. It can detect cancer and other diseases in radiology images which a physician might not see. It can recognize signs of depression and suicidality in patients’ voices, through changes in speech too subtle for a doctor to notice. It can predict disease processes days, weeks, even years before they become clinically evident. There’s a scene in the recent film Don’t Look Up where a smartphone CEO tells Leonardo DiCaprio that his phone data suggests he has a few colon polyps that should get checked out—it may sound funny now but could become a reality of AI in the future.
Freeing Healthcare Professionals To Focus On Care
Many providers and physicians are currently burdened by electronic health records (EHR) systems that often require them to spend more than half their time clicking and typing, instead of giving their attention to patients and families at bedside. (Disclosure: My company has invested in an AI solution for EHRs, and there are many companies focused on this issue already.) AI could automate and streamline much of this onscreen work, as it already has in other industries. It could potentially reduce “alarm fatigue” by filtering alerts so that healthcare teams are only interrupted by those upon which they have to act. And it could present the ever-growing amount of information in EHR systems as a summary in a physician’s preferred view, so they don’t need to flip through screens and scroll through records that aren’t relevant to the situation at hand.
Curbing Spiraling Costs
To anyone as familiar as we are at Inflect Health with both front-line medical care and emerging technologies, AI’s huge potential to rein in spiraling healthcare costs is obvious. In banking and retail, AI is already taken for granted—reducing transaction costs automating claims and adjudication processes, forecasting future surges in demand so that staff can be deployed where and when they will be most needed, and reducing overall administrative costs along with administrators’ time and frustration.
Why Healthcare Requires More Caution
Medicine, compared to other fields, calls for extra caution and patience in implementing change. We’re entrusted to care for people in literal life-or-death situations, and HIPAA rules on patient data privacy are rigid for good reason.
The physician-patient relationship is a sacred one that will, and must, always remain that way. My personal work with patients keeps me acutely aware that people want to be diagnosed by a physician, not a computer. On that note, I think AI has a bad reputation from the media and pop culture. The very term “artificial intelligence” provokes images of an all-knowing machine with a mind of its own. I think we need a new name for healthcare AI—perhaps “augmented intelligence”—that conveys to patients its proper role as an assistive technology for physicians, not a replacement.
In my next post I’ll detail some things AI definitely can’t do for healthcare. Meanwhile, there are two ways physicians can incorporate this technology in a way that keeps patients comfortable: First, emphasize improvements over perfection. AI won’t be flawless, but it can make healthcare better for patients. Second, and even more important, keep real patients and physicians involved in the development and testing of AI solutions. Engineers are smart at learning new fields and doing their research, but they can’t substitute for physicians who work with patients every day or the patients whose health we hope to better diagnose and treat using these promising new technologies.
Forbes Business Council is the foremost growth and networking organization for business owners and leaders. Do I qualify?