Photo: Camilo Jimenez/Unsplash
Biomedical engineers at RMIT University have built a smartphone feature that paramedics can use to immediately screen patients for stroke. In partnership with Brazil's São Paulo State University, RMIT University researchers developed an AI-powered tool for analysing facial symmetry and specific muscle movements, which are key signs of stroke. It is based on the Facial Action Coding System which categorises facial movements by the contraction or relaxation of facial muscles. The AI, in tandem with image processing tools, was tested on video recordings of facial expressions of 14 people with post-stroke and 11 healthy people. Based on findings published in the journal, Computer Methods and Programs in Biomedicine, the AI tool achieved 82% accuracy in detecting stroke "in seconds." The research team is now seeking collaborations with healthcare providers to turn their AI-driven smartphone feature into a mobile application. They are also considering expanding its use to detect other neurological conditions affecting facial muscles.
Citing studies, Dinesh Kumar, an RMIT University professor who supervised the research, noted that 13% of stroke cases are missed in emergency departments and community hospitals, while 65% of cases are undiagnosed. Gender, race, and geographic location can also contribute to overlooking strokes, he added. "Given that many strokes occur at home and initial care is often provided by first responders in non-ideal conditions, there is an urgent need for real-time, user-friendly diagnostic tools." A similar innovation in mobile health was done in the United States in 2020 by Penn State University and Houston Methodist Hospital. Their machine learning-based tool also uses computational facial motion analysis, as well as natural language processing, to detect stroke-like symptoms, such as sagging muscles and slurred speech. Other AI-driven stroke risk assessment and detection capabilities are applied to brain scans, such as the recently approved NNS-SOT by Nunaps in South Korea and AICute by Chulalongkorn University researchers in Thailand. There are also mobile applications in Asia, such as the telemedicine app DrGo in Hong Kong and RhythmCam by the National Taiwan University Hospital, that have introduced a-fib detection functionalities.