Deep Medicine - Eric Topol
This is a curious book. I think I agree with all the examples and disagree with the conclusion.
AI can transform medicine. Agree. Most of the non-medical examples given are a little well worn, but still useful as an introduction to the field.
It misses out some of the more interesting technical details, but provides an overwhelming amount of medical information. I found myself skipping large chunks which I was simply not qualified to read.
The conclusion is that AI will let doctors spend more time with their patients. Provide more empathy and human understanding. That's where I disagree. I long for the day I can go to a computer, plug myself in, and get a dispassionate diagnosis and treatment plan.
My preference isn't to have a human being gently explain to me what drugs I need to take… I like doctors! But I don't like interacting with them at their work. I don't want to show my vulnerabilities (or my rude bits) to a stranger.
We're not there yet - obviously. And the majority of people prefer a personal interaction with another human - again, obviously.
But for some people, a chat bot or automated process is more desirable. Much more so.
That's where I think the conclusion is wrong. The promise of AI is that we will see less of our doctors, not more. That common ailments will be picked up before they're a problem and complex needs won't be subject to a tired and fallible human misunderstanding something of infinite complexity.
The book is a great clip through the current state of medical AI. True, it is a little too credulous of some companies claims, and it focuses a bit too much on the wealthy world. But it does provide a detailed look at what comes next.
You should also read my colleague Indra Joshi's review in The Lancet.