Employment of AI in blood pressure related products and research is currently based, in most cases, on the AI processing of the data obtained with the PPG (photoplethysmography) technology. Below is the overview of the most notable projects update and the relevant recent studies. Further, two more non-PPG based projects are presented.
Relevant Projects
Shen.AI
Shen.AI utilizes AI through their SDK (Software Development Kit) by MX Labs to provide precise and easy-to-use camera-based diagnostics of vital signs and wellness. The SDK connects to the camera of a mobile/desktop device. It employs real-time Computer Vision to isolate a human face in the video stream. By constructing and tracking a 3D model of the face, the SDK extracts a high-quality, dense signal of blood pulsations using remote photoplethysmography (rPPG). This dense signal allows the SDK to accurately determine the timings and shapes of all observed heartbeats. And it provides users with precise heart-related metrics. All computations occur locally on the end device/browser, showcasing Shen.AI’s commitment to privacy and real-time analysis without external processing. The AI algorithms are embedded within the SDK enable advanced signal processing and analysis. They contribute to the accurate measurement of vital signs. These include heart rate, interbeat intervals, heart rate variability, breathing rate, cardiac stress, and blood pressure.
CamBP
CamBP, Remote Blood Pressure Measurement System is a non-contact system that uses AI algorithms to measure blood pressure by filming a person’s forehead. It is a more comfortable alternative to traditional cuff-based methods. The Contactless Blood Pressure Estimation System from MDPI operates using a computer camera. The system captures and analyzes physiological signals from the user’s body, such as facial features and movements. By analyzing subtle changes in facial features or skin color variations, the system infers physiological parameters associated with blood pressure. By utilizing computer vision technology, the CamBP system can provide continuous and real-time blood pressure estimates without requiring direct physical contact with the user.
PPG Lifelight
The technology developed by Verhaert Masters in Innovation is called Lifelight, and it enables the measurement of absolute blood pressure directly from finger, wrist, or upper arm PPG signals. This innovative approach allows for long-term monitoring of hypertension patients without the need for a sphygmomanometer, offering new applications such as personalized drug dosage adjustments based on accurate blood pressure measurements. Lifelight captures blood pressure by detecting variations in color on the patient’s face known as ‘microblushes’ that correspond with each heartbeat. The tablet’s camera captures these color changes, and AI processes this data to provide information on blood pressure, heart rate, and respiration. This contactless technology aims to be accessible for patients to use at home through tablets and smartphones. The trials conducted in Hampshire and the Isle of Wight involved patients looking at a tablet screen, with their blood pressure captured in just 40 seconds.
Cardio X
A newly emerged start-up CardioX took a step in preventing Blood Pressure related diseases through a new way of monitoring and measuring. It developed a Blood Pressure estimation method that uses only ECG recorded by a wearable device, without needing any additional equipment. CardioX BP estimation technology is available via mobile app. Once installed on a smartphone, you need to import an ECG recorded by your wearable device (for example smartwatch), then your AI estimated Blood Pressure, heart rate, and major HRV metrics are displayed. The app allows to conveniently track user’s health data and to share it with a family or doctor.
AI in blood pressure: not only PPG
AI-Enhanced BP Wristband
Artificial Intelligence-Enhanced Blood Pressure Monitor Wristband: This project involves a wristband equipped with sensors based on piezoelectric nanogenerators and a deep learning model to predict blood pressure readings with high accuracy, according to the researchers at the Beihang University in Beijing, China. The Artificial Intelligence-Enhanced Blood Pressure Monitor Wristband works by using sensors based on piezoelectric nanogenerators to predict blood pressure readings accurately. These sensors generate power from mechanical stress and are integrated with a deep learning model that processes the data collected to predict blood pressure values. The device filters and segments raw ECG data, extracts features, and uses machine-learning algorithms to create predictive models for systolic BP (SBP), diastolic BP (DBP), and mean arterial pressure (MAP). This process ensures continuous and reliable blood pressure monitoring for users. The wristband provides accurate predictions with a loss value lower than 4 mmHg. By incorporating piezoelectric nanogenerators, the device operates without external power sources.
AI Korotkoff Sounds Algorithm
This study presents an automated blood pressure estimation algorithm using sequences of Korotkoff sounds, showcasing advancements in blood pressure estimation techniques
Recent Relevant Studies on AI in blood pressure
Assessment of Blood Pressure Using Only a Smartphone. This study explores the use of smartphones and machine learning for blood pressure assessment, emphasizing the importance of regular monitoring for early detection of cardiovascular issues.
Another interesting and highly relevant study is ‘Recent developments in machine learning modeling methods for hypertension treatment’ published by The Nature in January 2024. The study looks into various explainable AI techniques that have been proposed for different purposes of model interpretation in the recent years.
