A gimlet-eyed look at the role of computers in medicine.
Building on earlier fly-on-the-wall looks at modern healing (The Patient Will See You Now: The Future of Medicine Is in Your Hands, 2015, etc.), cardiologist Topol examines the pros and cons of putting artificial intelligence, database crunching, and the like into the service of doctors who may or may not appreciate the new powers gained and limits reached. In this, the question is one of building a body of testable data and using it wisely. As the author writes, “shallow evidence…leads to shallow medical practice, with plenty of misdiagnosis and unnecessary procedures.” The data is more abundant than the meaning derived from it—by most estimates, Topol writes, doctors have collectively absorbed perhaps 5 percent of the whole literature. AI is useful for plowing through that huge body of material and weeding out the inapplicable and unlikely. AI is not, however, yet up to the “outlandish expectations,” as he puts it, that some administrators—and, more to the point, cost-cutting insurers—are placing on it, from curing cancer to eliminating possible harm to patients to lessening workloads. To be sure, he notes, there are many places where an algorithm’s ability to “eat data” is most welcome, as with correlating a patient’s intake of fluids with his or her output of urine. Given that most Americans have their medical records scattered over many providers and insurers, it’s important that data be consolidated and put in the hands of consumers. Perhaps paradoxically, notes the author, “the only way it can be made secure is to be decentralized.” Another issue is the possible overreliance of doctors on data in the place of good practice, and Topol closes with the warning: "Machine medicine need not be our future.”
A cogent argument for a more humane—and human—medicine, assisted by technology but not driven by it.