AI improves diagnostics and so what?

Artificial intelligence is here to stay. My favorite statement in all this AI hype is to remind that answers are only helpful if we have questions which needs an answer. AI is massively used in diagnostics and risk prediction. So, what is the utility of the improved diagnostics if we have no clue how to treat those newly diagnosed cases.

Gao et al. concluded in their systematic review:

The pooled sensitivity was 0.96 (95% CI 0.93–1.00), and the pooled specificity was 0.95 (95% CI 0.91–0.99). […] The results supported that the AI-based systems had good accuracy in diagnosing osteoporosis.

The application of AI-based systems in osteoporosis diagnosis needs to be further confirmed by more prospective studies in multi-centers including more random samples from complete patient types.

It is very nice that AI-based diagnosis of osteoporosis is accurate. But what is the question an AI-based diagnostics system would answer? If we can diagnose osteoporosis more accurately with AI, how should this translate to population level health. That is the most important aspect. Under-diagnosis is not a big issue in orthopaedics and in medicine in general. Sure in some aspects but not in the same scale we face issues with overdiagnosis.

There is a very big controversy regarding screening and pharmacological treatment of osteoporosis. Highly recommended readings about this topic include Overdiagnosis of bone fragility in the quest to prevent hip fracture and The true cost of pharmacological disease prevention. So it is relevant to ask: with such a controversy in the treatment of osteoporosis, what is the added benefit AI-based diagnostics have to offer? Are we just entering an era of unprecedented overdiagnosis and overtreatment?

Leave a Reply

Your email address will not be published. Required fields are marked *