About This Project

Can we use AI and a smartphone to accurately monitor heart rate in real time? We believe machine learning can detect physiological signals from facial features by analyzing subtle color changes linked to blood flow. To test this, we will develop and validate a system against clinical-grade heart rate monitors, assessing accuracy across different conditions. If successful, this non-invasive technology could provide an accessible tool for cardiovascular health tracking.

Ask the Scientists

What is the context of this research?

Smartphones and cameras, enhanced by artificial intelligence (AI), are emerging as key tools in health monitoring, yet real-time, non-invasive heart rate tracking is underexplored. Wearable devices, while effective, are costly and inaccessible to many, and traditional medical assessments require specialized equipment, limiting their reach in underserved areas.

Existing smartphone-based systems lack validation under real-world conditions like diverse lighting, motion, and skin tones, reducing their reliability. At the same time, the global rise in cardiovascular diseases highlights an urgent need for affordable, scalable health tools.

This research addresses these gaps by investigating whether AI-powered smartphone cameras can deliver accurate heart rate monitoring. By analyzing the capabilities of current technologies and the unmet health needs of diverse users, this study aims to explore how innovative, non-contact solutions can bridge critical gaps in accessible healthcare.

What is the significance of this project?

This project is significant because it pioneers AI-driven heart rate detection using smartphone cameras, a technology not yet available in the market. With over 6 billion smartphone users worldwide, this innovation could provide an affordable and accessible alternative to traditional heart rate monitors, especially in underserved regions lacking medical equipment. By enabling home monitoring, this research may help detect early signs of cardiovascular issues, manage stress, and improve overall health outcomes.

What are the goals of the project?

This project aims to develop an AI-based system capable of accurately detecting heart rate from facial selfies using a smartphone camera. The research will begin by training a machine learning model on publicly available datasets containing facial videos with corresponding heart rate annotations. The model will be optimized to detect subtle facial color changes caused by blood flow, ensuring accuracy across diverse skin tones, lighting conditions, and facial angles. To validate its performance, It will be tested using at least 1,000 annotated videos, assessing its robustness and reliability. The AI model will then be integrated into a mobile app, enabling real-time heart rate monitoring. This research will evaluate the model’s accuracy against ground truth data from datasets, without testing it against wearable technology. Potential limitations include inaccuracies under extreme lighting or high facial movement, which will be refined through iterative development.