A.I. is helping people see into the future — literally
By Leila Nsangou
One in every three diabetics is predicted to develop diabetic retinopathy, which experts expect will become the global leading cause of blindness over the next 30 years. Nearly 20 percent of all diabetics live in India, where there’s limited access to screening and treatment.
In recent years, Google has been developing artificially intelligent algorithms that will help prevent diabetic retinopathy, a disease that can cause blindness without treatment.
“There’s a lot of people, especially in India, where everyone needs screening, and this kind of blindness is completely preventable,” says Lily Peng, a biomedical engineer and physician who works with Google.
Google is only one company among many working to install A.I. in the medical world. Researchers at Harvard, Oxford, and other institutions around the world have started similar projects.
As A.I. has become more prominent in our world, it has nestled its way into important parts of our everyday lives. Photo and facial recognition are constantly being used in our smartphones and social media platforms. Household helpers like Google Home and Amazon’s Alexa are popping up in more homes every day. Even driverless cars are beginning to debut on public roads.
The use of artificial intelligence in health care is also becoming more advanced.
Doctors are starting to use A.I. to detect anomalies in photos to diagnose illness and disease. Just as machine learning can be used to identify a cat in a picture, it can be used to point out an aneurysm in a CT scan.
After being shown a large number of aneurysms, artificially intelligent technology can find the flags of sickness on its own. In China, doctors are already using A.I. algorithms to look for lung cancer.
Dr. George Shih, the founder of MD.ai, as well as a physician and professor at Weill Cornell Graduate School of Medical Sciences, told Wired magazine that A.I. won’t completely take the place of trained doctors.
According to Shih, doctors do a lot, and illness identification is only a fraction of their tasks.
“There are dozens of other pathologies that we are still responsible for,” he said.
Shih has been working towards the use of A.I. not only for illness detection but also for other fields. He believes that this type of A.I. will truly change health care, “particularly in the developing world, where trained doctors aren’t as prevalent.”
The lack of doctors in developing countries means that many people don’t have access to healthcare. This is where A.I. will make an impact profoundly.
Four years ago, Google started its work on an A.I. technology that can identify diabetic retinopathy, the leading cause of adult blindness. Initially, the project was one of Google’s “20 percent” ideas, creative projects that are dedicated to workers’ own personal passions. Later, they dedicated full-time resources to the project.
Google researchers partnered with hospitals in the Indian cities of Aravind and Sankara. They had doctors identify aneurysms, hemorrhages, and other indicators of diabetic blindness in retinal photos. After being reviewed by doctors, the photos were inputted into their neural networks — technology designed to function like the webs of neurons in our brains.
Deep learning, the same type of algorithm which is used to identify items in pictures, is used to find this sickness by looking closely at retinal photos. Through exposure to the labeled pictures, the A.I. has learned identify the anomalies itself.
Peng oversees the project. “We were able to take something core to Google — classifying cats and dogs and faces — and apply it to another sort of problem,” she said to Wired magazine.
These algorithms were mostly successful — right off the bat they had a precision and accuracy rate of 80 percent. If it continued this way, Google would help prevent blindness among diabetic patients.
At a 2017 Wired business conference, project leader Peng said that the technology was already being used in some eye hospitals in India.
Google is now working with Aravind Eye Care System, a network of eye hospitals that have been around for about forty years. The hospital network partnered with Google to help them to develop their algorithms.
According to Peng, India is one place of many around the world where there’s a lack of ophthalmologists. This means a lot of diabetic people don’t get the screening they need to work against diabetic retinopathy.
“This kind of blindness is completely preventable, but because people can’t get screened, half suffer vision loss before they’re detected,” she said. “One of the promises of this technology is being able to make healthcare more accessible.”
