Globally, the World Health Organization estimates that approximately 1.3 billion people live with some form of visual impairment. Although most of them are over 55 years old, there are 19 million children in the world with vision problems. Also according to the WHO, between 70 and 80% can be prevented or cured if they are diagnosed in time. If this does not happen, the consequences affect physical and social development: 80% of learning problems in children are caused by visual impairments.
But although there is a will to solve it there are two obstacles. One of them is age. “We are talking about very young children, less than 3 years old or with developmental problems,” Marta Ortín, an engineer and postdoctoral researcher at the University of Zaragoza explains. “And they cannot be made a regular exploration, that is to say show a letter or a drawing. And the other problem is that in the pediatric protocol, vision control does not enter until 3 years and at that time it may already be late.”
To try to solve this, Ortín, together with Victoria Pueyo, an ophthalmologist at Miguel Servet University Hospital, has created DIVE Medical, a startup that has developed the DIVE software (Devices for a Comprehensive Visual Examination) that allows monitoring and tracking the look of each eye of the patient while following stimuli specifically designed to test different aspects of visual function. This alone is already an important step forward in how exams were done until very recently.
“Before, the ophthalmologist stood in front of the patient,” Ortín adds in a telephone conversation, “and taught him some stimuli, designed decades ago. The results were very little objective and could not be repeated. However, the device we have designed is a screen where we show specifically designed drawings and add eye tracking, which means that the system knows where the patient is looking at the screen. From this we can establish a much more precise analysis.”
It is at this moment when artificial intelligence arrives. DIVE Medical joined Huawei and the Aragón Health Research Institute (IIS Aragón) have created the next step: Track AI. The idea is simple: use the ability to analyze and learn artificial intelligence to discover patterns not only in the eye but also in the eye movement and then give the results and, at the same time, the diagnosis.
“The analysis by itself,” Ortín points out, “gives us very valuable information but that must be interpreted. Thanks to AI we can go further. We analyze the look patterns to not only extract the information but also to indicate the probability that it has a certain pathology.”
The interesting thing about this collaboration is that much of the system is already tested. Track AI is based on Google TensorFlow, an open source platform, and Huawei HiAi artificial intelligence, present in the Kirin 980 microchip. Once a child’s eye test is done, a smartphone with the Kirin 980 (Mate 20 or the new P30) receives the data in real time and, without the need of a Wi-Fi network, processes it instantly. This requires a huge “library” with information from thousands of patients. And this has already begun to be done. In five countries on three continents (China, Mexico, United Arab Emirates, Spain and the United Kingdom), several research centers are working with thousands of children, collecting a large amount of data.
The fact of creating this library in different countries is a very good measure. It is about AI learning different colors of eyes, skin, face configuration, pupil size and visibility. There are numerous examples of artificial intelligences that have certain problems to differentiate male or female faces or even have prejudices towards different races. And that is something that in this case is being avoided from the beginning.
The idea is that the whole system is working later this year. The objective is to implement it in 2020, which will help health professionals diagnose and treat eye conditions at an early stage.
“We would like it to be used in places where it is difficult to turn to an ophthalmologist,” concludes Ortín. “In fact we are studying in the future to use it also for other ailments, not only ophthalmological, that can be detected with an ocular analysis.”
And this would be a very important advance. According to the American Academy of Ophthalmology, diabetes, high blood pressure, autoimmune diseases, sexually transmitted diseases and various types of tumors are among the diseases that can be detected during an eye exam.
2844 Petunia Way, Birmingham AL 35215