Artificial intelligence AI can now diagnose lung, breast, brain, skin, and even the cervix.
Factors such as the time that medical professionals have and their level of expertise can not be left out of the game. To address this, researchers from his Computer Science and Artificial Intelligence lab MIT (CSAIL) have developed a machine learning system that can decide whether to make a prediction or entrust it to a specialist.
The most important thing is that system AI can customize when and how often to address a person, based on the availability, experience, and scope of their partner's practice. For example, in a busy hospital, AI technology can call for human help in diagnosis only when absolutely necessary.
The researchers trained the system in multiple tasks, including examining chest X-rays to diagnose conditions such as a collapsed lung. When asked to diagnose cardiomegaly (an enlarged heart), the human-AI hybrid model he did it 8% better than the medical professionals themselves could.
«There are many obstacles that prevent it from being complete automation in a clinical setting, including matters of trust and liability," he says David Sontag, lead author of the paper presented by the CSAIL team at the International Conference on Machine Learning.
“We hope that the AI technology method of diagnosis will inspire its professionals engineering to become more creative in integrating real-time human experience into their algorithms. "
The next step for researchers is to test an artificial intelligence system that works and assigns to many experts at once. For example, he will be able to work with different radiologists who are more experienced with different patient populations.
The team also believes that artificial intelligence could also have an impact on content monitoring, as it can detect offensive text and pictures. As social media companies struggle to eliminate misinformation and hatred, a tool like AI technology, in addition to diagnosis, could ease the burden. work those who deal with it.