The Technology Behind Medios AI
The World's First Validated, Offline On-Device Artificial Intelligence Assistant For Screening Diabetic Retinopathy

Deep learning

The Medios AI relies on deep learning, the same technology that is behind self-driving cars or state-of-the-art speech recognition. It consists of a new coding paradigm that relies on large quantities of example data instead of hard-coded rules to teach a computer how to solve a task. This new approach led to significant breakthroughs in various fields in recent years.

Mimicking the brain

The algorithms behind the Medios AI are called Artificial Neural Networks. They consist of billions of simple mathematical operations applied on top of each other, computing a DR diagnosis out of the pixel values of a retinal image. Most of these operations have coefficients that have been set through exposing the neural network to tens of thousands of retinal images with their respective diagnosis. The name "neural network" comes from the similarity of this architecture with the way human neurons work.