The magical world of AI in medicine

Happy new year! In the first post of 2020 let's look at the most common prediction I could find in the digital health: AI will disrupt medical practice in 2020.

AI has long (i.e. for over 50 years) been known to be the next big disruption in medicine. Recent advances in so called "Deep Learning" algorithms that have fundamentally changed how computers interact with, in particular, images. We've seen Deep Learning algorithms brought the focus on AI to the fore in various technologies like social network recommendation algorithms and self driving cars. It feels magical but is it just a cognitive bias? Wanting things to work well often makes us think they perform better than they do.

When compared to other technologies that have drastically changed medicine (for the better) without disrupting it, X-Ray imaging, keyhole surgery, antibiotics and others, the magnitude of the impact, even at full potential, of deep learning algorithms, is not nearly as big. If in the years to come, deep learning can predictably work, it will be adopted and will change medicine for the better, but it would not disrupt it.

We are around the peak of inflated expectations in the Gartner Hype Cycle and all talk of AI or deep learning disruption of medicine are based on unfounded extrapolations based on incremental improvements, or in other word, hype.

Further reading: Interpreting technology hype