AI-powered health and wellness tools: Personalising medical care at your fingertips

With an iPad’s front camera, artificial intelligence and sitting still for just 30 seconds, Vitals, an AI-powered app, can tell your vital signs by simply scanning the colour changes in your face.

Vitals was developed by Panoptic.AI, a Hong Kong-based healthtech company founded in August 2022. The health and wellness monitoring app can identify up to 15 health indicators, including your breathing rate, blood pressure and oxygen saturation, which can help track current lifestyle conditions and detect any potential health risks down the line.

As the colours in your face are affected by blood flow, signals that only show these changes are tracked, which can also filter out “blind spots” such as beards and tattoos. Next, the signals are sent to the company through the cloud, while any personal identifiable information is kept back on the user’s device.

Kyle Wong, CEO and co-founder of the start-up, says the product’s idea stems from their previous projects involving temperature screening and thermal imaging technology in large-scale areas, such as border control points and government facilities.

During the COVID-19 pandemic, the company’s team realised that it was challenging to identify asymptomatic patients who did not show signs of fever or had taken medication that lowers their temperature, said Wong.

“We were doing a lot of research about using a camera, trying to find what other features we can measure from the person,” Wong said.

“That led to the idea of what we have now, which is by using a regular camera, and we're talking about the camera of your smartphone, your everyday, off-the-shelf device, we're able to measure these biomarkers,” he said.

Artificial intelligence is developing in Hong Kong’s health technology industry as it transforms health and well-being services into a personalised and self-manageable tool. 

The rise of artificial intelligence in digital wellness tools has also led more people to self-manage their health, according to a World Health Organization report published in 2021.

“You'll see that our technology has evolved a lot from these large-scale health screening systems to now a more personal version that's available on your smartphone,” Wong said.

Statista shows Hong Kong’s digital treatment and care market has grown by about 26% since 2020.  The data includes devices and apps that use connected biometric sensors or function as a health management tool. Forecasts show revenue is expected to grow to US$ 99.32 million (around HK$773 million) by 2028, another 22% increase from this year.

Hong Kong still faces a lack of resources needed to further develop AI technology. Out of 216 AI-related companies surveyed, 49% of them found it difficult to recruit technical talent due to high operating costs and not having the relevant skills, according to a report by the Hong Kong Productivity Council New Industrialisation this year.

To support technological advancement in Hong Kong, Chief Executive John Lee Ka-chiu first unveiled the Innovation & Technology Development  Blueprint in his 2022 Policy Address, including policies aimed at investing and transforming Hong Kong into an international hub in I&T in various fields, including health technology, artificial intelligence and data science for the next five to ten years.

Some universities have tried adopting artificial intelligence in health monitoring and management.

AI to Advance Well-being and Society (AI-WiSe), a joint-uni research platform by the University of Hong Kong and the University of Cambridge, developed a health management app called “UMeAir” that focuses on air quality data around the city.

Han Yang, 33, a postdoctoral researcher at the University of Hong Kong working on the project, says that artificial intelligence can be used to collect large and complex data sets.

The ongoing project aims to use machine learning to create an air pollution monitoring system, where it can generate personalised travel routes for users to avoid highly polluted areas in real-time.

Currently, there are 18 air monitoring stations in Hong Kong, with one of them located in a rural area called Grass Island, in the northeastern part of the region.

“This kind of AI approach actually is more efficient compared to traditional and physical air quality models. It can also be adapted to other cities where the data sets are valuable,” said Yang.

As the project also includes using personal health data from asthmatic patients, ensuring data privacy and security is important, says Yang.

“So we all know that these AI models are like black boxes. But, we don’t know the underlying process in terms of how and why this AI model gives you this kind of prediction,” Yang added.

Although there is no legislation in Hong Kong specifically governing artificial intelligence, the Office of the Privacy Commissioner for Personal Data published a guideline in 2021 to help organisations understand and comply with the requirements under the Personal Data (Privacy) Ordinance when developing and using AI.

To build transparency and trust in the emerging technology, some companies and researchers have opted for explainable AI , a set of processes and methods that can show AI’s decision-making so that human users can understand it. 

“We always have to find the balance between whether we can explain what our model is doing with the accuracy of the model,” said Wong. “We can obviously put in a lot of computational resources, train some very complex models and get a very high accuracy, but have no idea how they're doing it.”

“So especially in the medical field, that is really crucial,”  he said.

However, Kelvin Siu, CEO and co-founder of Innospire Technology Limited, says the decision to use explainable AI for consumers is not always necessary.

The tech startup created a visual assistance app called WeVoice using AI-text recognition to guide the visually impaired and elderly with poor eyesight, such as reading texts and banknotes. To make the app hands-free, it was further developed into a pair of smart glasses called WeVoice Glasses.

“Bridging that knowledge gap depends on the user interface designed by the engineer,” he said.“It is mostly the user interface that determines whether the logic behind AI or generative AI should be known.”

In practice, explaining the process of AI is also difficult because of the complexity of deep learning models, added Siu.

“Because we do computer vision, we need to control the input and output. So we need to be careful with the data inputted into the AI model,” Siu added.

A 2022 research study surveying around 20 organisations showed that explaining the machine learning process in real time is technically challenging.

Organisations found it difficult to search for certain data points quickly because the complex nature of some AI models causes it to give out multiple possible solutions but no absolute best answer. Currently, most AI-generated results are also based on rough estimates or the closest example from a different data category in the AI system, according to the study.

As the healthcare industry in Hong Kong continues to shift towards artificial intelligence, what future role it will play in changing human behaviour and the healthcare system is yet to be known.

“Technology is neutral. So, it can be used to do good things and also bad things,” said Yang. “It’s up to humans to use this tool. We need to make sure we design and deploy this AI for social good.”

Wong says that artificial intelligence is already changing the health technology and wellness industry, helping even clinicians in some very specific tasks.

“I definitely still stand on the side of AI still being used as a tool rather than a replacement for medical practitioners and doctors,” said Wong. “However, there's this ongoing quote that says, ‘AI won't replace doctors, but doctors who don't use AI will be replaced’."