A new groundbreaking app presented at the Society of NeuroInterventional Surgery’s 20th Annual Meeting, reveals how machine learning can be used to detect physical signs of stroke.
Machine learning can detect the physical signs of stroke
The study included 240 stroke patients from four metropolitan stroke centres.
Researchers recorded videos of the patients within 72 hours of the onset of stroke symptoms. They tested well-known physical signs of stroke, such as facial asymmetry, arm weakness, and slurred speech.
Using machine learning to optimise health
In order to evaluate facial asymmetry, the researchers employed machine learning techniques to analyse 68 facial landmark points. They then utilised a smartphone’s built-in 3D accelerometer, gyroscope, and magnetometer data in order to test arm weakness.
Mel-frequency cepstral coefficients were used to detect changes in speech, converting sound waves into images to compare patterns between standard and slurred speech.
The app was then evaluated by using neurologists’ reports and brain scan data.
The app demonstrated high sensitivity and specificity in diagnosing stroke accurately in most cases.
App could help to revolutionise stroke care with early detection
Dr Radoslav Raychev, a vascular and interventional neurologist from UCLA’s David Geffen School of Medicine and collaborator on this study has expressed their excitement about the potential impact of this app and machine learning technology could have for stroke care.
This revolutionary stroke detection app that uses machine learning has shown promise in aiding early identification of stroke symptoms, which could potentially save lives and improve care.
