AI Technology and Medical Applications
I started reading the Article that Mrs. LaClair gave us. I got down to the article section "Autonomous AI" of the technology article and found the authors argument about AI interesting. For one, the author seemed weary of artificial intelligence which does come on good merit. I found this piece on AI interesting and wanted to know more. I came across this article shorty after titled "The AI that can predict if you're going to die - but no one knows how" (The Telegraph, 12 Nov 2019). This article intrigued me and I decided to read the article. Turns out AI might be the thing we have been missing.
According to The Telegraph, an AI system from a US Healthcare center can compile millions of sets of data from patients and can form accurate predictions about the patients lifespan. The algorithm analyzed 1.77 million ECG results from 400,000 patients and predicted with 85% accuracy whether a person would die from health conditions or not (The Telegraph, 12 Nov 2019). After the scientists and doctors realized that their powerful AI tool can predict death they sorted through the data themselves to see if they could find what the AI system was picking up on, they couldn't.
Doctors can utilize this AI to predict which patients are going to pass away, but that's not the biggest plus. This program is picking out a pattern in the data but doctors can't seem to find out what it is yet. This AI and other AI systems alike can sort through mass amounts of historical data, which would take doctors years, in seconds. These AI systems can help vastly improve healthcare by spotting problems and irregularities that can not be seen by doctors. There might be problems with these machines, including patient confidentiality leaks but overall as they become smarter and more secure these AI programs could be a huge asset to studying diseases, cold and flu strains and overall healthcare.
" Facebook Engineering Facebook to open-source AI hardware design ." 2015.
According to The Telegraph, an AI system from a US Healthcare center can compile millions of sets of data from patients and can form accurate predictions about the patients lifespan. The algorithm analyzed 1.77 million ECG results from 400,000 patients and predicted with 85% accuracy whether a person would die from health conditions or not (The Telegraph, 12 Nov 2019). After the scientists and doctors realized that their powerful AI tool can predict death they sorted through the data themselves to see if they could find what the AI system was picking up on, they couldn't.
 |
| "Facebook Engineering..." |
Doctors can utilize this AI to predict which patients are going to pass away, but that's not the biggest plus. This program is picking out a pattern in the data but doctors can't seem to find out what it is yet. This AI and other AI systems alike can sort through mass amounts of historical data, which would take doctors years, in seconds. These AI systems can help vastly improve healthcare by spotting problems and irregularities that can not be seen by doctors. There might be problems with these machines, including patient confidentiality leaks but overall as they become smarter and more secure these AI programs could be a huge asset to studying diseases, cold and flu strains and overall healthcare.
Hoggins, Tom. “The AI That Can Predict If You're Going to Die – but No One Knows How.” The Telegraph, Telegraph Media Group, 12 Nov. 2019, www.telegraph.co.uk/technology/2019/11/12/ai-can-predict-going-die-no-one-knows/.
Dickinson, Kevin. “Runaway Automation.” Big Think, Big Think, 5 Oct. 2018, bigthink.com/technology-innovation/fear-future-technology?rebelltitem=3#rebelltitem3.
Facebook, 10 Dec. 2015, images.app.goo.gl/ECxa2qSjL2QzeWz96.
Accessed 15 Nov. 2019.
Accessed 15 Nov. 2019.
The article you found is interesting. Was there any information in the article to suggest whether the doctors could use this information to help save the lives of the patients identified? How helpful is the information if it doesn't help save lives?
ReplyDeleteIt is really interesting that the doctors couldn't figure out a pattern in the information.
The doctors at the moment are not using this program yet, it is in the testing phase. Because they can't find the pattern that the program is using, right now there is really no way to tell what the doctors could do better or change.
Delete