Doctors, overall, are a very smart bunch, but they can be resistant to change. The most famous example is probably the 19th-century surgeon who refused to wash his hands when moving from the morgue to the delivery ward, spreading as yet undiscovered germs and causing the death of infants. Hungarian physician Ignaz Semmelweis, who collected data to claim that soap and water could save lives, was ridiculed and ostracized.
Today, we live in more enlightened times, and medical practice is generally supported by evidence – but are we always getting the right evidence to drive change? For example, there are signs that putting artificial intelligence into clinical use could even save lives. As we reported in “AI helps radiologists detect breast cancer in real-world trials,” radiologists who chose to use an image-classifying AI to help detect breast cancer Picked up one additional case per 1000 people tested. On health care systems, the impact could be huge.
Does this mean we should encourage doctors to take off their scrubs and take over the machines? Far away from. While large language model AI systems like ChatGPT can be successful in multiple choice medical trials, they perform less well on conversational diagnosis (see “AI chatbots fail to diagnose patients by talking to them”). A physician with a good bedside manner and a listening ear is still important.
We must be bolder in testing medical AI systems in real-world settings
Instead, we can draw two conclusions from these studies. The first is that we should be careful in using the general term “artificial intelligence.” Although the two systems we report on share an underlying neural network technology, image classification is a very different task to text generation, and in the latter the risk of AI producing plausible but incorrect results is much higher. In other words, not all AI is created equal.
The second conclusion is that we should be more adventurous in testing medical AI systems in real-world settings rather than just in a laboratory or simulation. The breast cancer study, by giving radiologists control over when to use AI, shows that it can be a useful tool. By attempting to obtain more such evidence, lives could be saved, just as Semmelweis, who is now considered a medical hero, did.
Subject:
(TagstoTranslate)Medical Technology(T)AI(T)Healthcare
