Tracking health metrics from a selfie? Face scan-based health checkup, a gimmick or reliable?
Recently, I came across a technology that blew my mind away.
My initial thoughts were that it was a gimmick. How could someone measure BP or SpO2, just from a phone’s front camera.
It scratched that part of my brain.
As I came across a lot of such services, I began my research. I had to dive deeper into this technology. And my findings were quite interesting summarized in this blog.
At the heart of this seemingly magical process lies a technology called photoplethysmography (PPG) — a term familiar in medical circles but now undergoing a reinvention through AI and computer vision.
So, let’s explore my findings.
Understanding PPG Technology
The advent of photoplethysmography (PPG) has revolutionized the potential for non-invasive health monitoring. It’s an old school method already popularly used in many medical devices.
But with the growth of mobile devices, internet and AI, started the rise of rPPG — remote photoplethysmography that is bringing this technology to the end consumer.
With just a front-facing camera, rPPG enables the estimation of multiple vital signs, including
- heart rate (HR),
- respiratory rate (RR),
- blood oxygen (SpO2),
- and even blood pressure (BP).
But how does this technology even work?
rPPG measures subtle changes in light absorption and reflection on the skin’s surface. This correlates with blood volume changes beneath the skin.
These variations are invisible to the naked eye but can be captured via video and processed using advanced AI algorithms to extract vital health data.
How It Works:
- Video Capture: A standard camera records the face, typically for 30–60 seconds.
- Signal Processing: Algorithms isolate specific color channels (red, green, blue) to detect pulsatile blood flow or a waveform.
- Vital Derivation: ML models trained on clinical datasets map these signals to physiological metrics.
There’s already one common application for PPG.
It’s in our smart watches to measure our heart rate and blood oxygen. That’s PPG as well!
There’s a lot of correlation between PPG and rPPG technology as highlighted by this study: Pulse Rate Variability Analysis Using Remote Photoplethysmography Signals — PMC
The AI Advantage
PPG technology was already getting used in certain closed room medicine applications. But with emergence of AI and especially complex computer vision models, a lot of things are changing.
Traditional PPG methods required clinical-grade equipment and ideal conditions. With AI, several limitations were overcomed:
- Signal Filtering: Noise caused by movement, lighting, or ambient conditions is reduced.
- Multi-Vital Estimation: Simultaneous calculation of multiple parameters from a single video.
- Demographic Adjustments: Algorithms trained on diverse datasets allow for skin tone, age, and gender corrections.
- Predictive Analytics: AI enhances diagnostics by detecting patterns not immediately visible in raw data.
How is the accuracy of these calculations?
- Blood Pressure (BP)
Recent research has validated that systolic and diastolic pressures can be estimated using rPPG with minimal deviation compared to cuff-based devices.
A study published in Nature Biomedical Engineering reports a mean absolute error (MAE) of 6.4 mmHg for systolic pressure, which is well within the FDA’s allowable limit of ±10 mmHg.
- Blood Glucose Estimation
While still in its experimental phases, the correlation between vascular micro-dynamics and glucose levels is promising.
Early trials have shown a regression coefficient (R²) of 0.85, suggesting strong predictive accuracy when calibrated against invasive glucose meters.
- Parasympathetic Waves
By analyzing heart rate variability (HRV), AI can infer autonomic nervous system activity, providing insights into stress levels and overall wellness. This application is particularly relevant in mental health assessments and corporate wellness programs. - Respiratory-Related Changes
Applications in respiratory health are particularly impactful for monitoring chronic conditions like asthma or post-COVID recovery
CVD and Diabetes is a leading cause of death in India and all across the world, and it has been shown that constant monitoring of blood pressure and glucose values would solve that issue.
Comparison between manual measuring and rPPG based technology:
Manual: Varies from how’s the reading is taken.
rPPG: A lot of this noise can be smoothened out.
Manual: Manual effort as a lot of devices will be needed.
rPPG: Calculates a lot of parameters and uses AI to adjust intelligently.
Manual: Lot more costly and time consuming.
rPPG: Very cost effective. Just a small video selfie and you are done for the day.
Applications of Face Scan
Face scan technology is transforming industries like insurance, healthcare, and wellness with innovative applications. In insurance, it simplifies claim processing, enhances underwriting through real-time health metrics, and detects fraud using biometrics. Wellness tracking integrated into policies fosters healthier behavior and better engagement.
In healthcare, it enables non-invasive diagnostics, continuous chronic disease management, and real-time vitals for telehealth consultations. Mental health applications use it for stress and emotional monitoring, while the wellness industry employs it for personalized health assessments and fitness tracking, empowering proactive health management.
Finally, is Face Scan a Gimmick or does it actually work? What did my analysis say?
Validation: Separating Fact from Gimmick
A study published in National Library of Medicine well summarized the correlation and disproved my theory of face scan technology being just a gimmick. Here are some details from that study:
The study compared readings taken from ECG with a rPPG system to measure vitals such as RRI and HRV metrics. There was a clear correlation found between the values measured from both the methods.
Not just this study, but multiple other research have proved that rPPG method is highly practical and usable, with only restriction being its fine-tuning as per the factors such as demographics, region, environment, etc.
Hence, AI plays a crucial role in bringing in further personalization to rPPG measurements, making them more accurate and reliable while replacing them with manual methods.
Ethical and Practical Considerations
In a recent webinar on rPPG technology and Data Privacy that I had attended by Binah, an interesting topic was discussed. The speaker was talking about how rPPG technology can seriously be dangerous for user’s privacy, as a lot of details can be extracted from a person’s face.
This information can be used to determine a lot of parameters on top of the vitals, such as age, gender, diseases, etc.
For e.g., if an AI system identifies someone as having a higher risk of cardiovascular disease based on their facial scan, that information could inadvertently influence insurance premiums or job opportunities.
Hence, it’s important that companies safeguard their users against such misuses and implement strictest data privacy policies.
Which applications I tried out?
Vista:
Vista by Medista leverages face scan technology in its unique way. It is highly simple to use and available through their application or through a web-based link. Vista provided me AI-based suggestions as well from my report.
Activ Health
Activ Health is another app available, that measures your vitals from face video. They calculate same number of parameters as Vista and is being offered by Aditya Birla Health.
Eka Care
Eka Care allowed me to calculate Heart Rate using just the camera. So, it was a close contact-based implementation of PPG.
Happy You
Happy You is another application I tried out that was available for free, providing a unique UI and other features.
Closing Thoughts
I have always been a proponent of AI and technology in improving human lives. Hence, bringing out such technologies in the open is really great way to improve health monitoring and help people stay on top of their health.
Selfie-based health monitoring, underpinned by rPPG and AI, is not just a tech gimmick — it’s a breakthrough with the potential to redefine how we monitor our health. While challenges remain, its applications are vast, and its accuracy continues to improve.
As researchers, innovators or users, we must remain vigilant in ensuring these tools are accessible, reliable, and equitable for all. The journey of face scan-based health monitoring is just beginning, and its future looks very bright.
