Physicians as a whole are a very smart group, but they can be resistant to change. The most famous example is probably the 19th century surgeon who refused to wash his hands when going from the mortuary to the delivery room, spreading undiscovered microorganisms that led to the infant's death. Hungarian doctor Ignaz Semmelweis collected statistics to argue that soap and water can save lives, but was ridiculed and ostracized.
Today, we live in more enlightened times and medical practices are generally supported by evidence, but do we always have the right evidence to make changes? For example, do we always have the right evidence to make changes? There are also indications that bringing it into clinical practice could save lives. As we report in “AI Helps Radiologists Detect Breast Cancer During Physical Exams,” radiologists who choose to use image classification AI to help detect breast cancer are detected one extra case per 1,000 people. Across the health system, the impact can be significant.
Does that mean we should encourage doctors to ditch the scrubs and let the machines do it? Far from it. AI systems with large language models like ChatGPT can pass multiple-choice medical tests, but do poorly at conversational diagnosis (see “AI Chatbot Fails Conversational Diagnosis of Patients”). ). Physicians with a good bedside manner and a listening ear remain essential for now.
We should be bolder when testing medical AI systems in real-world settings
Instead, there are two conclusions that can be drawn from these studies. The first is that we need to be careful in using the general term “artificial intelligence.” Although the two systems we report share underlying neural network technology, image classification is a completely different task than text generation, and the latter risks the AI spitting out plausible but inaccurate results. much higher. In other words, not all AIs are created equal.
The second conclusion is that we need to be more bold in testing medical AI systems in real-world settings, not just in the lab or in simulations. The breast cancer study showed that AI can be a useful tool, giving radiologists control over when to use it. This effort to obtain more evidence could save lives, as it did for Semmelweis, who is now considered a medical hero.
topic:
Source: www.newscientist.com
