It isn’t that often we get articles about AI / machine learning / LLMs in health in the NZ mainstream press:
2 Likes
Nice to see. And they’re right about the environmental impact. Otherwise, mostly Naah.
This is another variant on AI as deus ex machina.
I’d suggest that there are several things wrong with this view of AI driving precision medicine:
- 'Precision medicine has largely been an abject failure.
- AI in its current incarnation (or any reasonable extension) would be rather unlikely to deliver on the vision—even if precision medicine weren’t fatally flawed.
- The focus is wrong. This is the most important point.
Let’s look at (1) and (3) briefly.
There’s lots wrong with precision medicine. First, it has been promised for over two decades but pretty much hasn’t delivered. Second, it struggles to accommodate epigenetic factors. Third, it fails to accommodate somatic mutation which is particularly troublesome in tumours: if you sequence one tumour cell, this often tells you little about the totality of tumour cells.
But the main failing is that the entire focus is wrong. Health is increasingly about non-communicable diseases—and most of these are potently determined by environmental factors, especially how entire communities are dosed with mostly good things (like vegetables, vaccines, clean water and fresh air) or mostly bad things (like alcohol, salt, nicotine, high-energy-density food, opiates, meth and so on).
Hoping for some future AI ambulance at the foot of the cliff to fix things with imagined precision neglects the epidemiology of health and disease.
Fail.
My 2c, Dr Jo.
1 Like