Eirini Kateri, founder of remote monitoring startup Zenicam, demonstrates her product in a room with … [+]
A UK startup is trying to measure pulse, respiratory rate and even blood pressure via video calls to reduce the need for patients to travel to the doctor’s office.
Zenicam founder Eirini Kateri hopes her artificial intelligence software will provide a rich set of real-time data, helping doctors make decisions remotely and ultimately improving access to healthcare in hard-to-reach areas. was achieved.
Its early results look promising. But experts warn that the technology will face some major hurdles before it reaches the real-life doctor’s office.
Kateri first worked on an AI video pulse monitor as part of a PhD program at the University of Southampton in England. She told me she had a “lightbulb moment” when a friend was sick and struggled to contact a doctor during the pandemic.
“I thought, can I? [get this information] remotely to help people who don’t have access to doctors, or to help when doctors are overwhelmed and overwhelmed?” she said.
Gemma Galdón-Clavell, founder and CEO of the AI security organization Eticas Foundation, told me that Zenicam was “certainly ambitious” and described the startup’s focus on capital as “laudable.”
But she is skeptical that the technology can succeed in a space filled with technical challenges.
“These systems generally do not work. We’ve seen time and time again how facial recognition and emotion recognition technologies fail most of the time,” she said. “And Zenicam is trying to assess something even more nuanced: vital signs.”
Prejudice is a problem that Kateri hopes her approach can solve. Medical AI software can be prone to bias if it is trained on videos and images of people who look relatively similar.
It’s a problem found in medicine, where research and technology tend to be biased toward white people.
In the UK, medical professionals still miss and misidentify symptoms that appear differently in skin tones, sometimes with devastating results. Earlier this year, a government-commissioned report criticized medical equipment not working properly on black people.
Kateri believes she has found a way to ensure Zenicam is accurate regardless of a patient’s ethnicity or appearance. It’s trained on both real data and “synthetic data,” which it says can fill in gaps in existing data.
This is some kind of “fake data” based on real data that Zenicam has already received, she told me. You can calculate a wide range of different scenarios by “adding different motion or lighting or changing skin tone, scale or color based on existing data”.
“I could have had a video of someone not wearing glasses,” she said. “I could artificially add occlusions to the face or mask certain areas to cover scenarios where someone was wearing them.”
To work in the medical space, Zenicam will need to provide evidence that it can record vital signs in different environments, said Galdón-Clavell, who advises the United Nations and the EU on applied ethics and responsible AI. .
She will have to deal with the fact that most patients won’t be in a quiet, perfectly lit room, nor will they always have a stable Internet connection.
By training her AI models on these variations, Katerini says she can accurately extract pulses and other vital signs, regardless of people’s physical appearance or the environment they’re calling from.
But Galdón-Clavell urged caution with the use of synthetic data, which risks “exacerbating existing problems”.
“You first have to have a good understanding of real-world data and be able to assess what’s missing from the existing dataset,” she told me. “Synthetic data should be able to compensate for what is missing.”
This is “a growing specialty in this area” that requires expertise and serious supervision to work correctly, she added.
However, Kateri told me that her results have been promising so far. “For pulse we are accurate to within one beat per minute, for respiration to within two breaths per minute and for blood pressure for diastolic and systolic we have an error of up to 5 mmHG in all physical representations,” she said.
Measuring blood oxygen levels and temperature were “more challenging,” but the Zenicam was currently about 70% accurate.
The startup will need to provide full scientific evidence to support these claims if it wants to put Zenicam into clinical use.
Since the software is designed to help healthcare professionals make more informed decisions during a video call, “it needs to be validated as a medical device,” Kateri said.
Right now, her team is integrating AI technology with video conferencing systems “to show exactly how it will work and measure accuracy for regulatory bodies.”
She thinks this will take at least a year and a half. In the meantime, it hopes to conduct device trials in parallel to validate the technology.
If it works, Kateri says she hopes to eventually be able to take the product to countries where access to a doctor is rarer than in the UK.
“I’m most excited about the kind of potential the solution has to be used around the world, especially in underserved countries,” she told me. “Where we can help everyone, regardless of their background, to get access and fair access to health care.”