Strokes are one of the leading causes of death in the U.S., with someone dying of a stroke every 3 minutes. Each year, around 800,000 Americans experience a stroke.
A team of researchers have developed a new smartphone app, called FAST.AI, which could help people more quickly and efficiently identify the signs of stroke, allowing them to receive emergency care in a timely fashion and potentially improve outcomes. Findings from a preliminary study of FAST.AI will be presented at the American Stroke Association International Stroke Conference.
FAST.AI is powered by a machine learning algorithm trained to detect various outward symptoms of a stroke. For example, the app is trained to detect changes such as changes to facial symmetry or changes in a person’s speech, which are all key signs of a stroke. By collecting a simple video of a person, the app can detect these various changes.
To validate the app, researchers tested it on about 270 different people who had been diagnosed with an acute stroke in the past 72 hours. The team studied patients at four different major stroke centers in Bulgaria.
Overall, researchers found that the app could do the following:
- In all patients, the app accurately detected changes to a person’s facial symmetry.
- In about two thirds of patients, the app correctly detected signs of arm weakness.
- Showed potential for detecting slurred speech (this part of the app still needs to be validated).
Researchers hope that their tool could allow people suffering from an acute stroke to get timely treatment, which can make all the different in someone’s overall outcome. For example, it’s recommended that medication designed to break up clots be given to patients within just three hours of experiencing a stroke. The sooner treatment is administered, the fewer brain cells that die.
Currently, the app is still in development and not widely available for clinical use.