Google AI Predicts Heart Disease and Stroke From Retinal Images


The latest applies machine learning to retinal images to identify the risk factors of cardiovascular disease.

Google and Verily's scientists used machine learning to analyze a medical dataset of almost 300,000 patients, as per the report.

Comparing the image of the fundus in two patients, one of whom within the next 5 years suffered a heart attack, Google's algorithm learned to determine the risk with an accuracy of 70%. Google has taken that reality and combined it with deep learning algorithms to take the diagnostic potential to a new level. As part of this, researchers trained deep learning algorithms using data on 284,335 patients.

According to the researchers, the algorithm that this device uses is quicker than traditional predictor tests using blood tests. They can identify patterns in otherwise disparate data sets, in this case potentially helping to identify indicators of cardiovascular problems. In biology the rear interior wall of the eye is full of blood vessels that reflect the body's overall health. The system can also use retinal images to predict risk factors, which Google says includes things like the patient's gender and age.

More news: FCC chairman backs SpaceX plan for internet satellites

In a paper published today (via The Verge), Verily and the Google AI teams detail their work analyzing blood vessels at the back of the eye to predict risk factors, like blood pressure and smoking, associated with cardiovascular disease. "This performance approaches the accuracy of other CV risk calculators that require a blood draw to measure cholesterol", the blog post noted.

Sounds superficial? Well, it's a tested and proven method backed by Google's advanced artificial intelligence (AI) and machine learning in a project led by Alphabet's health-tech subsidiary Verily. In other words, it's not yet ready for clinical testing, but it's a promising start for non-invasive evaluation of cardiovascular health. These techniques helped the company to generate a heatmap which basically shows which pixels were the most important for a predicting a specific CV risk factor. Explaining how the algorithm is making its prediction gives the doctor more confidence in the algorithm itself.

Google also used some attention techniques to find out how the algorithm was making its prediction.

The algorithm that the researchers produced can even help predict the occurrence of a future major cardiovascular event, said Michael McConnell, Head of Cardiovascular Health Innovations at Verily.