Technology
Combining neuroscience, psychology, physiology and Deep Learning to produce an Affective AI engine that ultimately become the key provider under the universal health coverage (UHC) principles.
In the intersection between Affective Computing and Artificial Intelligence.
BitDoctors AI starts by capturing images of the subject using any conventional video camera including those found in a smartphone.
BitDoctor AI automatically detects and tracks your face identifying key regions of interest (ROIs)
Key biometrics taken will be encrypted and record into the blockchain for ultimate protection. No Image is taken during the process besides the information of the flow of the melanin and haemoglobin. This method provide the ultimate autonomous power while providing key information to the network.
A comprehensive and structured approach to encrypting medical data and incorporating it into a blockchain system. It covers key aspects such as encryption, data structuring, hashing, transaction submission, consensus mechanisms, access control, decryption, auditability, and compliance with regulations.
--------------------------------------------------------------------------------------------------------------------
Blood Flow Data Extraction
Human skin is translucent. Light and its respective wavelengths are reflected at different layers below the skin and can be used to reveal blood flow information in the human face. This information is captured by and contained in conventional video images. BitDoctor AI extracts this information and sends it securely up to the cloud to be processed by BitDoctor’s AI, our cloud-based Affective AI engine.
Signal Processing and Deep Learning
Extracted facial blood flow data is sent up to the cloud where our AI Doctor engine applies advanced signal processing and Deep Learning AI models to predict physiological and psychological affects.
An advanced machine learning algorithm to create computational models that predict reference systolic, diastolic, and pulse pressure from facial blood flow data. We used 70% of our data set to train these models and 15% of our data set to test them. The remaining 15% of the sample was used to validate model performance.
We found that our models predicted blood pressure with a measurement bias±SD of 0.39±7.30 mm Hg for systolic pressure, −0.20±6.00 mm Hg for diastolic pressure, and 0.52±6.42 mm Hg for pulse pressure, respectively. The results in normotensive adults fall within 5±8 mm Hg of reference measurements. Future work will determine whether these models meet the clinically accepted accuracy threshold of 5±8 mm Hg when tested on a full range of blood pressures according to international accuracy standards.
Smartphones equipped with BitDoctor AI imaging technology may meet these requirements. BitDoctor AI imaging technology is a recently developed variant of remote photoplethysmography for imaging blood flow patterns from video of the face. Video-based photoplethysmography capitalizes on the following facts. First, because of the translucent nature of facial epidermis, ambient light can penetrate the epidermis and reach the tissue below, with some of it reflected back out of the skin. Second, the digital optical sensors in smartphones are highly sensitive and thus can capture re-emitted light and its small attenuations. Third, the quantity of hemoglobin protein in the blood and melanin pigment in the skin determines the color of light that is reflected back out of the skin. Each has a different color signature, so it is possible to separate re-emitted light containing mostly hemoglobin information from light containing melanin information based on the differential absorbance characteristics of these 2 light-absorbing proteins.
In BitDoctor AI imaging technology, light from the visible spectrum travels beneath the skin surface and is re-emitted before being captured by the camera sensor. BitDoctor AI imaging technology capitalizes on subtle changes in skin color from the difference in re-emitted light between hemoglobin and melanin chromophores to detect blood flow pulsation in the cardiovascular system. The process of BitDoctor AI imaging technology involves (1) capturing video of the face using a conventional camera, (2) extracting spatiotemporal images of hemoglobin concentration from the bitplanes of the red, green, and blue image channels using advanced machine learning, (3) processing the hemoglobin signal from 17 different regions of interest, (4) extracting features from these signals, and (5) using a blood pressure prediction model trained with advanced machine learning algorithms to indicate blood pressure from these signals.
Results
Results processed by BitDoctor AI Imaging engine are then sent back to your device for display and further analysis.
The accuracy measurements have been trained and validated. The computational models against already well-established scientific instruments found in labs and clinics. Accuracy and validity of the measurements have been tested at both BitDoctor.ai and the university labs.
Last updated
