Depth is all about discovering a new dimension. When you add width to length and height, you create volume, which just happens to be a synonym for “book.” I’ve just finished Eric Topol’s latest tome, “Deep Medicine: How Artificial Intelligence Can Make Healthcare Human Again,” and was fascinated by its multidimensional narrative.
Although artificial intelligence (AI) is more than human, Topol took great pains to make “Deep Medicine” a book of human dimensions. It mimics the direction the author wants medicine to take: to become more comprehensive, personal, effective, and humane as AI improves. He goes out of his way to reveal numerous sides of himself (patient, caregiver, husband, son-in-law, father, doctor, and researcher) as a way to embody health care’s holistic future.
Topol begins by talking about his difficult experience following knee surgery. He was put on a course of vigorous physical therapy that included some anguished episodes on a stationary bike. “The pain was well beyond the reach of oxycodone,” he writes. “A month later, the knee was purple, very swollen, profoundly stiff, and unbending.” The problem was that Topol had developed arthrofibrosis, the formation of excessive scar tissue after injury or trauma (like knee surgery) that restricts a joint’s motion. His orthopedist didn’t treat it properly, though his kind physical therapist, who halted the exertions and texted regularly to inquire about “our knee,” did.
Topol’s attitude toward the orthopedist — who after replacing Topol’s knee advised him to ask his internist for anti-depression medications — is not one of fraternal understanding but of angry hurt. “The robotic response of my doctor to my distress exemplifies the deficient component of care. Sure, the operation was done expertly, but that’s only the technical component.” He says that, yes, the operation left him emotionally depressed but “the problem was that I was in severe pain and had Tin Man immobility. The orthopedist’s lack of compassion was palpable.”
AI “could have predicted that my experience after the surgery would be complicated. A full literature review, provided that experienced physical therapists such as the woman I eventually found shared their data, might have indicated that I needed a special, bespoke PT protocol,” Topol writes, adding that a “virtual medical assistant, residing in my smartphone or my bedroom, could warn me, the patient, directly of the high risk of arthrofibrosis that a standard course of physical therapy posed.”
The majority of “Deep Medicine” is taken up with a skilled researcher’s thorough catalogue of AI studies, AI startups, and the relative failures and successes of technical ventures up until the book’s publication date (which is to say, right now). There’s an impressive catalogue of stories about how machine learning has done some incredible diagnostic work, about AI ingesting thousands of medical articles, of technology analyzing Topol’s own gut microbiome. In this the reader encounters chilling sentences such as, “Machine learning of mammography images from more than 1,000 patients, coupled with biopsy results indicating risk of cancer, showed that more than 30 percent of breast surgeries could be avoided.”
There are so many mini-case studies here that it can be overwhelming. I don’t think my case study fatigue was entirely because I’m a non-medical person. I think it’s because there’s just too much take in at once. (Let me cite the Nabokov Defense here: “Curiously enough, one cannot read a book: one can only reread it. A good reader, a major reader, an active and creative reader is a re-reader.”)
Topol has clearly read all the AI and machine learning material a human can read and met everyone in the space worth meeting. The book covers AI drug discovery, AI mental health (Instagram filters apparently say much about one’s mental state), and the “deep liabilities” AI brings. “All too commonly we ascribe the capability of machines to ‘read’ scans or slides,” Topol writes, “when they really can’t read. Machines lack of understanding cannot be emphasized enough. Recognition is not understanding; there is zero context…”
When talking about how AI can help with palliative care, Topol recreates the final stretch of the life of his beloved nonagenarian father-in-law, John. Healthy all his life, John is quickly overtaken by a variety of horrible symptoms, a few trips to the emergency department, and then the eventual discovery that he has end-stage liver disease which, Topol writes, “didn’t make sense, since his drinking history was moderate at worst.”
AI, he says, could have given everyone involved a much clearer understanding of what was happening to his father-in-law as the man’s life closed down. Topol talks about the dying algorithm, a digital neural network 18 layers thick based on the electronic health records of nearly 160,000 people. It was, he writes, “able to predict the time until death on a test population of 40,000 patient records with remarkable accuracy.” He mentions that Google and a trio of medical centers are now working with 47 billion data points to predict “whether a patient would die, length of [hospital] stay, unexpected hospital admission, and final discharge diagnoses.”
In addition to the numerical considerations of the end of John’s life, Topol also paints an emotional picture of it — tests, medical bafflement, wild vomiting of blood, hands-on healing from John’s daughter and granddaughter, hospice preparation, a miraculous moment of resurrection, and finally death. And he makes John’s death personal. “Against the wishes of his nurses, I packaged him up and took him in front of the hospital on a beautiful fall afternoon. We trekked down the sidewalk and up a little hill in front of the hospital; the wind brought out the wonderful aroma of the nearby eucalyptus tree. We were talking, and we both started to cry. I think for him it was about the joy of being alive to see his family. John had been my adopted father for the past twenty years, since my father had died, and we’d been very close throughout the nearly forty years we had known each other.”
Unlike some physicians and researchers, Topol doesn’t have a God complex. If anything, he has a human complex. He understands that physicians are deeply limited, and that AI could one day radically extend their medical capabilities. Right now, even the best doctors operate within the parameters of human cognition. “There are about 10,000 human diseases, and there’s not a doctor who could recall any significant fraction of them. If doctors can’t remember a possible diagnosis when making up a differential, then they will diagnose according to the possibilities that are mentally ‘available’ to them, and an error can result.”
Topol wants us to understand and accept the limitations of the medical mind — and then use powerful tools like AI and machine learning to transcend them. But he knows we’re not there yet. He talks about, for instance, the impossibility of a human reading the 2 million peer-reviewed biomedicine papers published each year, and then smiles at ads suggesting that with IBM Watson by one’s side, a doctor could “read 5,000 studies a day and still see patients.” Watson, he says, does ingest abstracts, but it doesn’t transform all that data into a structured database that would be useful to a working doctor.
“Deep Medicine” is not a perfect book. There are moments when Topol’s descriptions are less than elegant (“He seems a bit like a Larry David personality with curbed enthusiasm, something of a curmudgeon,”) and he has an unfortunate tendency to provide stage directions at the end of chapters. The armchair editor in me wishes he had balanced his narrative elements more the way Robert Pirsig had balanced alternating bursts of philosophic, travel, and autobiographical narrative in “Zen and the Art of Motorcycle Maintenance.”
But the glitches don’t detract from Topol’s vision of health care’s future, in which the asymmetry of today’s medicine evolves into something more substantial, in which patients and physicians really understand each other.
Topol gives all readers — doctors, patients, health care administrators, machine-learning experts, basically anyone who cracks the spine of the book — a visceral sense of what the holistic medical future could be. He’s clearly fed up with the presto tempo of contemporary medicine, the dehumanizing distractions of electronic health records, the perilous burnout rate among doctors, and the inefficiencies of health care and its outsized costs. He hopes that new technologies, accompanied by a new mindset, will slow the health care process and make it possible for everyone to get to know each other.
“Deep Medicine” offers the mindset needed for the change Topol seeks. The experience of reading this book is, in a way, the opposite of AI. You don’t just ingest this narrative and spit out the patterns picked up from the text. You read, and you think, and you feel, and you wonder — and you do it all over again every few pages. Naturally.
Ken Gordon is the content, conversation, and community strategist at EPAM Continuum, a global innovation design firm.
This is the future of medicine, at the very least in the developing world where technology can also fill in for the doctor. Marginalized communities need medical help that is affordable and can scale, something that technology does well. Great review Ken.
Old fashioned intelligence and common sense is probably better!
I found that by taking a personal history that the cause of psychiatric symptoms lept out at us, and the symptoms abated.
With psychotic patients it was more complicated: their symptoms dated from infancy when the right cerebral hemisphere was very active, but the left hadn’t yet developed linguistic ability, and recovery required patient/therapist cooperation to nurture the neglected inner infant.
Using Persig as an example of how to improve writing is painfully funny.
Comments are closed.