Unpleasant surprise: Artificial intelligence divides patients into poor and rich

Models with artificial intelligence may recommend different treatments for the same medical condition based solely on the patient’s socio-economic and demographic characteristics, researchers warn.

This turned out to be after the necessary patients were tested in nine health -speaking artificial intelligence (AI) health -speaking models, which were asked how they should be treated in thousands of different situations in the emergency room.

Despite the identical clinical details, AI models sometimes change decisions based on the personal characteristics of patients, affecting the priority of care, diagnostic tests, treatment approach and evaluation of mental health, according to researchers in Nature Medicine magazine.

For example, sophisticated diagnostic tests such as computed tomography or nuclear magnetic resonance have been more often recommended for high-income patients, while low-income patients were more often advised not to undergo additional tests.

Researchers have found that problems are observed in both patented artificial notelect models and open source models.

« By identifying where these models can bring bias, we can work to improve their design, strengthen the supervision and build systems that guarantee that patients remain at the heart of safe and effective care, » added co-author Dr. Eyal Clang, also from the Ikan school.