Anmol Madan’s The Machines Will See You Now takes on a big question: what happens when artificial intelligence moves from assisting doctors to making meaningful health care decisions itself? Madan calls this future “Autonomous Health,” and he builds his case by tracing the progression from telemedicine and digital health to AI systems capable of diagnosis, treatment recommendations, personalization, and ongoing care management. The comparison to self-driving cars gives the book an accessible starting point, but Madan quickly shows why medicine is a much more complicated problem. A machine caring for a person has to understand far more than data. It has to account for context, behavior, communication, risk, and eventually the sensory information that an experienced clinician picks up almost instinctively. The book’s three-part structure moves logically from that vision into the technology required to make it possible and then into the practical questions surrounding implementation, regulation, equity, and jobs.

What made the book especially engaging to me was Madan’s willingness to move between technical explanation and personal experience. He draws on his background in machine learning, digital health, telemedicine, and health care startups, so many of the ideas come with examples of what happens when AI leaves the research lab and has to work for actual consumers and organizations. Chapters on automated diagnosis, personalization, hybrid care, and passive sensors make the technology understandable without turning the book into a textbook. I found the discussion of personalization particularly compelling because Madan connects familiar recommendation systems to a much higher-stakes goal: finding the intervention, message, or care experience most likely to improve an individual person’s health. Throughout the book, he also keeps returning to an important constraint. Medical AI needs more than impressive language skills. It needs reliable information about the person it’s treating and ways to recognize when its own capabilities aren’t enough.

The later chapters give the book more weight because Madan turns from what the technology can do to what organizations and society will have to do with it. His sections on building AI products, regulating automated care, reducing inequity, confronting algorithmic bias, and retraining the health care workforce turn the book into something closer to an operating guide for the next stage of digital medicine. His perspective is clearly optimistic, and he often writes as though significant health care automation is a question of when. That confidence gives the book momentum and makes its predictions interesting to wrestle with. It also means some readers will probably be more cautious than Madan about the pace of adoption or the ability of AI to reproduce parts of clinical judgment and human interaction. Still, the book earns its optimism by spending substantial time on validation, safety, regulation, bias, workforce disruption, and the practical difficulty of moving from a promising model to a trustworthy health care system.

I’d recommend this book to people who are going to have some responsibility for deciding how AI gets used in health care. Health system executives, digital health founders, product managers, clinicians working with new technologies, investors, policy professionals, and students entering health technology will find plenty to think about here. It’s particularly useful for someone who already understands that AI will influence medicine but wants a clearer picture of how all the pieces fit together, from data and clinical decision-making to economics, regulation, and staffing. Madan has written a forward-looking guide that treats autonomous health care as an organizational and social transformation as much as a technological one. Even readers who don’t share every prediction are likely to come away with a sharper set of questions about who will build these systems, how they should be tested, what doctors will do alongside them, and what kind of health care experience they could create.

Pages: 272 | ASIN: 1421455056