Polyclinic patients to trial AI tool that reads heart rate and blood pressure with just a phone camera

Straits Times

20-06-2025

By analysing a 30-second video of a user's face, the tool is able to detect levels of blood flow in specific facial regions such as the forehead, cheeks and nose. PHOTO: SINGHEALTH POLYCLINICS
Polyclinic patients to trial AI tool that reads heart rate and blood pressure with just a phone camera
SINGAPORE - An artificial intelligence tool that allows people to measure their heart rate and blood pressure with just a phone camera will be tested by a group of polyclinic patients in the coming months.
DeepHealthVision, developed by South Korean digital health company injewelme, will be trialled here in collaboration with SingHealth Polyclinics (SHP).
By analysing a 30-second video of a user's face, the tool can detect levels of blood flow in specific facial regions such as the forehead, cheeks and nose, producing accurate readings of both heart rate and blood pressure without the need for external medical devices such as blood pressure monitors and oximeters.
'For example, someone who has high blood pressure or heart rate commonly presents with flushed skin,' said injewelme's chief business development officer Nathan Cheong on June 20, at the 10th Asia Pacific Primary Care Research Conference 2025 held at Academia at the Singapore General Hospital Campus.
The accuracy of measurement is not affected by skin colour, he added.
Apart from phones, patients can also use the cameras in other devices such as laptops and tablets.
Clinical Associate Professor Tan Ngiap Chuan, director of research and Primary Care Research Institute at SHP, said this would make healthcare monitoring more accessible and convenient by eliminating the need for expensive medical devices and frequent clinic visits.
A check by The Straits Times found that digital blood pressure monitors can cost anywhere between $50 and $300.
'This system also facilitates continuous remote health monitoring, allowing healthcare professionals to detect health issues before they become serious, enable earlier intervention, and potentially reducing hospitalisations,' said Prof Tan.
Mr Cheong said there are also plans to integrate injewelme's AI prediction model DeepHealthNet into DeepHealthVision, which would allow healthcare professionals to access patients' vital signs so that they can make more informed clinical decisions, instead of relying only on limited information collected during visits to the doctor.
More than 100,000 people in South Korea have tested out DeepHealthVision, while around 30 healthcare professionals in Singapore were able to try their hand at it during a recent live demonstration .
During the demonstration here, the accuracy of DeepHealthVision's heart rate and blood pressure readings was around 96 per cent, when compared with readings taken by Omron blood pressure monitors or Samsung Galaxy smartwatches, said Mr Cheong.
Prof Tan said around 60 polyclinic patients will be involved in the pilot test of this tool in the coming months, which would include those with and without hypertension, and others who are interested in knowing more about their own vascular health.
He said: 'We want to find out whether local patients are confident in using the technology, whether they accept it, and if they have any concerns. Because if they have concerns, they might not use it on a regular basis.'
The pilot test is expected to run for at least a year, during which the data collected will be analysed and subjected to peer review.
If the trial is successful, DeepHealthVision may be integrated into healthcare apps such as Health Buddy in the future.
The team is also working to enable blood glucose monitoring and skin analysis down the line.
In May, SHP also signed memorandums of understanding with two other South Korean companies, YoungAnd and Monit.
SHP is collaborating with YoungAnd on Brain Health Playground, an interactive platform designed for patients to play games that support cognitive function. SHP will also trial a tool developed by Monit to assess diaper contamination levels with the use of sensors and an AI-driven pattern learning algorithm.
Join ST's WhatsApp Channel and get the latest news and must-reads.