Skip links

From Selfie to Diagnosis: Detect Heart Disease by Snapping Pictures

We all take them. Soon selfies might actually keep us healthy.

AI, Selfies, and the Future of Medicine 

Based on how quickly AI-based machines are advancing, the future is very exciting. The healthcare industry is on the cutting edge, and experts are encouraging the adoption of wearables to improve diagnosis and treatment. First, there were smartwatches that could monitor heart health. Now, it seems that taking a simple selfie can help detect heart disease.

Kotanidis and Antoniades recently published this idea in the European Heart Journal. Their research was inspired by the findings of Zheng et al., who studied the feasibility of using deep learning models to detect coronary artery disease (CAD) based on facial photos of patients. Zheng et al. used a convolutional neural network model, and their cross-sectional study included patients from nine Chinese sites undergoing coronary angiography or computed tomography angiography.

With this resourceful and affordable approach, patients in underserved regions around the globe could simply send selfies from anywhere and receive a proper analysis. Although the research is still in its early stages, AI experts are confident this will be a breakthrough for AI and disease diagnosis.

The Selfie Phenomenon

Focusing on selfies is an easy way for science and technology to transform regular habits into valuable tools for improving overall health. Not everyone can visit a physician who will monitor their cardiovascular health. However, we can all find a moment to take a selfie. 

“Selfie” is one of the most used words of the 21st Century. The most popular word in 2013, it has only grown in usage ever since. Nathan Hope first introduced the term in an internet journal in 2002.

Twenty-four billion selfies were posted on Google’s servers in 2015, and now that number is up to 93 million selfies PER DAY. On average, a given person will take eight selfies every 24 hours.

An Example of a banner with a button

Get the book for free

How Selfies Can Detect CAD

Utilizing selfies, machine learning algorithms can pick out certain facial features that indicate disease conditions like CAD. Some include the appearance of wrinkles, thinning or grey hair, and small, yellow deposits of cholesterol under the skin, usually around the eyelids (xanthelasmata). Also, deposits of fat and cholesterol appear as hazy white, opaque blue, or grey rings in the outer edges of the cornea (arcus corneae).

Prof. Zheng and his team studied the role of facial recognition models in the detection of heart diseases. In their research, they recruited 5,796 patients from eight Chinese hospitals between July 2017 and March 2019. These patients were already undergoing imaging procedures focused on their blood vessels. The subjects were divided into two groups for better data acquisition and processing. 

They trained their deep learning algorithm with data generated by four facial photos taken with digital cameras: one frontal, two profiles, and one view of the top of the head. Research nurses recorded their medical history, lifestyle, and socioeconomic status information. To establish credence, radiologists reviewed angiograms to assess each subject’s degree of heart disease based on narrowed blood vessels’ total concentration and location.

The research concluded that their deep learning algorithm outperformed existing methods by accurately predicting heart disease risk in 80% of the cases. However, the fact that this research model was only tested on Chinese people means it did not consider factors unique to other races and geographical locations. 


1. Selfies, already a huge part of our daily lives, can help us detect coronary artery disease. 

2. The selfie as a diagnostic tool is a concept still in its early stages. However, if widely utilized, it could be a significant breakthrough that could facilitate better health even to underserved and remote regions.

3. As always, with artificial intelligence, the possibilities are endless.

Leave a comment

This website uses cookies to improve your web experience.