1. Face Video Screening Technology
Face video screening technology in a nutshell
- Face videos can be used to detect vital signs and blood biomarkers and assess a person’s health status and risk of developing diseases.
- The underlying technology is called Photoplethysmography (PPG). It leverages RGB light reflection patterns from a person’s face to detect changes in the skin and blood vessels.
- The technology can detect blood pressure, heart rate, heart rate variability, respiratory rate, stress levels, oxygen saturation, and, soon, even cholesterol and sugar levels (glycated hemoglobin). This information can be used to assess cardiovascular health (e.g. arrhythmias risk, stroke risk, hypertension risk, diabetes risk), respiratory health (e.g. sleep apnea), and mental health (e.g. stress and anxiety levels).
- The accuracy varies depending on the biomarker/ disease and the length of the face video. However, the technology showed relevant results, achieving accuracy levels higher than 90%.
What is Photoplethysmography (PPG) - the technology behind face video health screenings?
PPG is a non-invasive optical technique that measures blood volume changes in tissues. It operates on the principle that blood absorbs and reflects light differently depending on its concentration and oxygenation level. By emitting light into the skin and detecting the reflected or transmitted light, PPG sensors capture variations in blood volume during each cardiac cycle.
How does the technology work?
RGB light reflection is a technique that involves shining red, green, and blue light onto a person's face and measuring the amount of light that is reflected. By analyzing the patterns of reflection, machine learning algorithms can detect changes in the skin and blood vessels that may be indicative of certain diseases. For example, changes in the amount of light reflected from the skin can indicate the presence of skin cancer, while changes in the blood vessels can indicate cardiovascular disease.
More in detail, PPG detection process can be broken down into the following steps:
- Light Emission: PPG begins with the emission of light, usually in the visible or near-infrared spectrum, into the tissue. Light-emitting diodes (LEDs) are commonly used as light sources.
- Tissue Interaction: The emitted light interacts with the tissue, penetrating the skin and reaching blood vessels beneath. Oxygenated blood absorbs more light than deoxygenated blood, resulting in variations in light intensity.
- Light Detection: Photodetectors, such as photodiodes or phototransistors, capture the light that is either reflected or transmitted through the tissue. These detectors convert the received light into electrical signals.
- Signal Processing: The electrical signals from the photodetectors undergo signal processing to extract the pulsatile component corresponding to cardiac activity. This component represents changes in blood volume with each heartbeat.
2. Detecting Risk of Diseases through Voice Recordings
What are voice biomarkers?
Voice biomarkers refer to specific patterns, characteristics, and changes in vocal signals that are associated with various physiological and pathological conditions. By analyzing these biomarkers, experts can gain insights into a person's health status, detect early signs of diseases, and monitor the progression or response to treatment.
How Does Voice Biomarker Analysis Work?
The main steps for voice biomarker analysis are the following:
- Voice Recording: The process begins with capturing voice recordings using digital devices or smartphones. These recordings can be obtained in controlled environments or through mobile applications.
- Feature Extraction: Advanced algorithms extract specific features from the voice recordings, such as pitch, volume, tempo, spectral characteristics, articulation, and emotional patterns. These features serve as the basis for further analysis.
- Machine Learning Analysis: Machine learning techniques are applied to the extracted features to identify patterns and correlations between vocal characteristics and specific diseases or health conditions. Models are trained using large datasets to classify and predict the signs or progression of diseases.
Diseases Detectable through Voice Biomarkers and Related Accuracy Levels
Diseases Detectable through Voice Biomarkers and Related Accuracy Levels:
- Parkinson's Disease: Voice biomarkers can aid in detecting early-stage Parkinson's disease. The vocal characteristics of individuals with Parkinson's exhibit distinct tremor, reduced vocal loudness, and changes in speech rhythm. Studies have shown promising results with accuracy levels ranging from approximately 80% to 95% in detecting Parkinson's disease using voice biomarkers.
- Alzheimer's Disease and Dementia: Voice analysis techniques can assist in the early detection and monitoring of Alzheimer's disease and other forms of dementia. Vocal features such as speech rate, pauses, and intonation patterns can exhibit changes associated with cognitive decline. While accuracy levels may vary, studies have reported accuracies ranging from around 75% to 95% in detecting Alzheimer's disease through voice biomarkers.
- Depression and Mental Health Disorders: Voice recordings can provide valuable insights into mental health conditions. Changes in vocal tone, pitch variability, and speech patterns have been correlated with depression and other mental health disorders. Accuracy levels for detecting depression using voice biomarkers have been reported in the range of approximately 70% to 90%.
- Respiratory Conditions: Voice analysis has shown potential in detecting respiratory conditions such as chronic obstructive pulmonary disease (COPD) and asthma. Abnormalities in vocal characteristics, such as increased breathiness or changes in airflow patterns, can be indicative of respiratory disorders. Accuracy levels for detecting respiratory conditions using voice biomarkers have been reported in the range of approximately 70% to 90% (e.g. COPD, Asthma, Covid). However, many conditions are still being explored, and further research is needed to establish their robustness.
3. Data Privacy in our App Beta
We use HIPAA compliant models and we don't store face video. For detailed information on our privacy policy please consult the following page at this link: https://app.termly.io/document/privacy-policy/15b9d38a-e10b-4cbf-974f-e393673f0e89