An AI that was trained at 1.77 million electrocardiograms is good at predicting who will die within a year. It can find heart problems even in people who doctors consider to be healthy.
Image by Gerd Altmann from Pixabay
- Be also sure to check out my other posts and follow me
@kralizec and subscribe to my Youtube channel at Kralizec Gaming Youtube Channel
Artificial intelligences are amazing, breathtaking and sometimes pretty scary. The last adjective perfectly fits a new medicinal AI that was trained by experts from a local health system in Pennsylvania and New Jersey called Geisinger Health System (GHS). This AI is pretty good at predicting which patients will die within a year. But there is a problem. Even its creators do not know how the AI achieves its results.
The experts fed the AI with a total of 1.77 million EKGs that came from roughly 400 thousand patients. Their goal was to have the AI learn to distinguish situations that would be a mark on future heart problems. The results are truly impressive. And a bit creepy. The trained AI is much better than any currently used methods to predict the health of patients.
The research lead – Brandon Fornwalt – says that they found no model that would be better than their AI. It can pretty reliably predict which of the patients will die within a day. And it can even detect problems in people who cardiologists said are healthy for their age. Fornwalt is convinced that their AI notices things humans are either incapable of noticing, ignore as noise, or think of as fluctuations within the norm. The AI could teach us which things are important for patients with heart problems that we have ignored for decades.
Funnily enough, this isn't the first death-predicting AI out there. Google specialists from the Californian Mountain View developed an AI last year that based on patients' history predicts how long will they stay in a hospital and sometimes even their deaths. And some other AIs diagnose serious health problems like heart problems or lung tumors better than their human colleagues.
The reason why I am writing about the Geisinger AI is that even its creators do not know how the AI works. That truly scares me. And it is also one of the reasons why some experts warn us from making rash decisions based on AI predictions. Yet, I do not agree with the experts. Just because we do not understand how the AI works doesn't necessarily mean they should not be trusted.
Sources:
- https://futurism.com/neoscope/ai-knows-when-youll-die-how
- https://www.newscientist.com/article/2222907-ai-can-predict-if-youll-die-soon-but-weve-no-idea-how-it-works/
- If you like the content I’m producing about science maybe you will like the content I produce about gaming as well! Be sure to check out my other posts!