Smartphone camera and flash could help people measure blood oxygen levels at home – ScienceDaily

First, stop and take a deep breath.

When we breathe, our lungs are filled with oxygen, which is distributed to our red blood cells to travel throughout our bodies. Our bodies need a lot of oxygen to function, and healthy people have at least 95% oxygen saturation all the time.

Conditions such as asthma or COVID-19 make it difficult for the body to absorb oxygen from the lungs. This drops oxygen saturations to 90% or less, which is an indication that medical attention is needed.

In the clinic, doctors monitor oxygen saturation using pulse oximeters — those clips you place on the tip of your finger or ear. But monitoring oxygen saturation at home several times a day can help patients monitor COVID symptoms, for example.

In a proof-of-principle study, researchers from the University of Washington and the University of California San Diego demonstrated that smartphones are able to detect blood oxygen saturation levels up to 70%. This is the lowest value a pulse oximeter should be able to measure, as recommended by the U.S. Food and Drug Administration.

This technique involves participants placing their finger on a camera and smartphone flash that uses a deep learning algorithm to decode oxygen levels in the blood. When the team administered a controlled mixture of nitrogen and oxygen to six people to artificially lower blood oxygen levels, the smartphone correctly predicted whether the person had low blood oxygen levels 80% of the time.

The team published these results on September 19 npj digital medicine

“Other smartphone apps have been developed that do this by asking people to hold their breath. But people feel very uncomfortable and have to breathe after a minute or so, before their blood oxygen levels drop enough to represent the full range,” said the co-lead author Jason Hoffman, a PhD student at the University of Washington at the Paul G. Allen School of Computer Science and Engineering.”With our test, we can collect 15 minutes of data from each subject. Our data shows that smartphones can perform well in the critical range.”

Another benefit of measuring blood oxygen levels on a smartphone is that almost everyone has one.

“This way you can get multiple measurements with your own device either at no cost or at low cost,” said co-author Dr. Matthew Thompson, professor of family medicine at the University of Washington School of Medicine. “In an ideal world, this information could be seamlessly transmitted to the doctor’s office. This would be really useful for telemedicine appointments or for triage nurses so they can determine if patients need to go to the emergency department or if they can continue to do so quickly. Rest in home and make an appointment with your primary care provider later.”

The team recruited six participants, ages 20 to 34. Three of them were identified as female, and three were identified as male. One participant was identified as African American, while the rest were identified as Caucasian.

To collect data to train and test the algorithm, the researchers asked each participant to wear a standard pulse oximeter with one finger and then place another finger on the same hand on a smartphone camera and flash. Each participant has the same setting on both hands simultaneously.

“The camera records a video: Every time your heart beats, new blood flows through the part that the flash lights up,” said senior author Edward Wang, who started the project as a doctoral student at the University of Washington studying electrical and computer engineering. Assistant Professor in the UCSD Design Laboratory and Department of Electrical and Computer Engineering.

“The camera records how much the blood absorbs light from the flash in each of the three color channels it measures: red, green and blue,” said Wang, who also directs the UC San Diego DigiHealth Laboratory. “Then we can enter these density measurements into our deep learning model.”

Each participant breathes in a controlled mixture of oxygen and nitrogen to slowly reduce oxygen levels. The process took about 15 minutes. For all six participants, the team took more than 10,000 readings of a blood oxygen level between 61% and 100%.

The researchers used data from four participants to train a deep-learning algorithm to pull oxygen levels in the blood. The rest of the data was used to validate the method and then tested to see how well it performed on the new subjects.

Co-lead author Varun Viswanath, a UW alumnus and now a doctoral student advised by Wang at UCSD, said. “Deep learning is a really useful technique here because it can see these really complex and subtle features and help you find patterns that you wouldn’t be able to see otherwise.”

The team hopes to continue this research by testing the algorithm on more people.

“One of our team members had thick nails on his fingers, which made it difficult for our algorithm to accurately determine blood oxygen levels,” Hoffman said. “If we expand this study to more subjects, we will likely see more people with corns and more people with different skin tones. And then we can have an algorithm with enough complexity to be able to better model all of these differences.”

But the researchers said this is a good first step toward developing biomedical devices aided by machine learning.

“It is very important to do a study like this,” Wang said. “Traditional medical devices are subject to rigorous testing. But computer science research is just in its infancy using machine learning to develop biomedical devices and we are all still learning. And by forcing ourselves to be tough, we are forcing ourselves to learn how to do things right” .

Additional co-authors are Xinyi Ding, a doctoral student at Southern Methodist University. Eric Larson, Assistant Professor of Computer Science at Southern Methodist University; Caiwei Tian, ​​who completed this research as an undergraduate at UW; and Shwetak Patel, UW Professor in both the Allen School and the Department of Electrical and Computer Engineering. This research was funded by the University of Washington. The researchers applied for a patent covering SpO2 classification systems and methods using smartphones (application number: 17/164,745).