Smartphone camera can detect diabetes mellitus with high accuracy
The number of people with diabetes continues to grow every year. Even 10 years ago, when affordable smartphones with digital cameras appeared, doctors assumed that they could be used to diagnose this disease. Now, a technique has been created that can determine the presence of diabetes with fairly high accuracy only with the help of a smartphone camera.
Diabetes can be asymptomatic for a long time. This complicates its diagnosis and subsequent treatment. One of the most popular non-invasive methods for detecting diabetes is photoplethysmography (PPG), in which light is sent to the tissue to detect changes in blood volume. This technique is often used to measure heart rate and oxygen levels. A team from the University of California, San Francisco (UCSF) has shown a working way to diagnose type 2 diabetes using a smartphone camera.
First, the researchers uploaded 2.6 million PPG records from 53,870 people with diagnosed diabetes to the database. Then, they developed an algorithm based on a neural network that can effectively distinguish people with diabetes from healthy people using biomarker data. During the experiment, it was found that the AI was able to correctly diagnose diabetes in 80% of patients. After the algorithm got access to the data of body mass index and age of people, the accuracy of diagnostics significantly increased.
"We have demonstrated that the efficiency of the algorithm is comparable to other widely used tests. This affordable smartphone-based tool can be used to identify people with an increased risk of diabetes and encourage them to seek medical help and take an inexpensive confirmation test," the study authors report.
Scientists see a huge potential in the new method of detecting diabetes since in the future it will be possible to monitor the state of health without having to go to a clinic. However, doctors have not yet announced the release date of the application for smartphones.