New AI Technology can Diagnose Humans with Cancer
By Jack Tendler
A group of London doctors have created artificial intelligence software to correctly diagnose humans with different types of cancer, but with problems such as being prone to computer hackers, it may not be worth trusting this technology. It is already a technology that some people don’t trust, but the benefits could certainly outweigh the negatives.
The technology uses a neural network which is a form of modern artificial intelligence (AI). Programmers feed it thousands of photos containing examples of cancer and then they give it just as many examples of healthy patients. Over time the machine learns what cancer looks like and what a healthy human looks like.
The technology becomes extremely accurate, and after time it becomes just as if not more accurate than a trained professional. It can identify skin cancer, predict when a heart will fail, and even look for kinds of brain cancer.
“Most dermatologists were outperformed by the CNN [Convolutional Neural Network],” the research team wrote in a paper published in the journal Annals of Oncology. “On average, human dermatologists accurately detected 86.6% of skin cancers from the images, compared to 95% for the CNN,” the Guardian reports.
On top of determining skin cancer, AI technology can actually predict when someone’s heart is about to fail.
“A team of doctors in London have trained AI to predict when the heart will fail…It could transform treatment in ophthalmology, dermatology, and Radiology,” the BBC reports.
Humans apparently can predict things like when a heart will fail around 80 percent of the time. While 80 percent may seem like a decent score on a math test, not being able to identify when someone’s heart will fail 20 percent of the time is not good at all. While the AI technology trained by London doctors isn’t perfect, the BBC says it “has greatly outperformed human cardiologists.”
While talking about the benefits of relying on AI technology, another important thing to realize is the price. The price for using AI technology is actually significantly cheaper then using a trained professional. In an interview with the BBC, healthcare tsar Sir John Bell talked about the price benefits of using this technology.
“There is about $2.97 billion spent on pathology services in the NHS(National Health Service),” said Bell. “You may be able to reduce that by 50 percent. AI may be the thing that saves the NHS”
The last benefit that this technology has over humans is fatigue. Humans get tired. A trained doctor can have years of experience but they will always be prone to this one major flaw. They can’t be available 24/7 without sacrificing accuracy. A machine can always be available and never take breaks. This surprisingly enough could be the most useful of the features as being available constantly greatly adds availability to the system.
With all of these benefits it may be easy to immediately love the idea of using AI in the medical field, but there are of course drawbacks too. As technology increases so do different forms of hacking.
While it is easy to guess that hackers wouldn’t focus on the medical field, that would be a dangerous assumption. If we were to rely on AI to diagnose cancer patients and the technology were to be hacked then thousands of lives would be completely ruined. People who don’t need treatment would get it and people who desperately need it could go without it.
According to the ITRC Data Breach Reports, “Medical and Healthcare entities made up 35.4 percent of data breach targets (276 breaches),” and that data was only from 2015.
While a three year difference may not seem like a lot, in the technology world that feels like a decade. The landscape of the technology field changes so commonly that so much can change just in one year.
According to Revision Legal, the amount of cyber attacks keeps going up with no sign of stopping.
“Cyber-attacks are happening in 2017 at double the rate of 2016,” Revision Legal reports. That would include double the amount in the medical field. And again keep in mind that that data is from 2015. This means the amount of medical breaches could be nearing the thousands.
Another thing to keep in mind is that if the AI technology were to be implemented nationally, it could add more incentive for computer hackers to attack hospitals and try and alter the AI machine.
It’s a huge risk. One of the biggest advantages humans have on these AI computers is that they can’t be affected by some computer hack.
If hospitals were to incorporate this technology they would have to potentially hire more programmers just to ensure that the technology is safe from outside harm.
The possible benefits of using new AI technology could potentially be life saving.
However, we must first get over the mental hurdle of allowing computers to make these life changing decisions for us.
We need to ensure that the technology is being monitored frequently to make sure there is no hack to either change the outcome of a scan, or to collect private data. If we could do this, then this technology could theoretically change the world.
