James Amihood, CEO of digital telehealth company Cardiokol, spoke to Health Tech World about voice analysis and how it can be used to detect atrial fibrillation.
What is atrial fibrillation and why is it so important to identify?
Atrial Fibrillation (AF) is a growing epidemic. It’s the most common heart rhythm disorder and a major health risk, as untreated AF is associated with cardioembolic complications, primarily stroke.
In recent years, 20 to 30 per cent of all ischemic strokes have been related to AF. Additional complications of AF include heart failure, heart attack, dementia, kidney failure, death and more.
These acute complications can be largely prevented by starting use of oral anticoagulants once AF is detected. Therefore, there is a strong need for early detection of AF in populations at-risk.
What makes conditions like atrial fibrillation difficult to catch with isolated tests?
AF onsets are often intermittent and asymptomatic, therefore very difficult to capture during tests.
Detecting asymptomatic, intermittent AF requires continuous or frequent and systematic testing over time.
At the moment, current methods for AF detection are either inaccessible to most of the older population at risk, because they’re unaffordable and age-unfriendly, or provide low detection yields (symptoms-based/short-term monitors).
What is a vocal biomarker?
Vocal biomarkers are medical signs extracted from the voice features. They can be used to detect health conditions in previously undiagnosed populations at risk, or to monitor chronic patients as part of their disease management.
Cardiokol’s first product, Voice-Assisted Arrhythmia Monitoring (VAAM), uses daily voice samples from which it extracts certain features, and then analyses them using AI and signal processing algorithms.
The product then flags chaotic heartbeats, indicating AF onset, once an episode or burden is identified.
What are the benefits for the patient on using voice analysis to identify atrial fibrillation?
The primary target population at risk of AF are over-65s. Current AF monitoring methods are not cost-effective or elderly-friendly over time , primarily for the high-risk, asymptomatic, undiagnosed older adults.
By using daily voice samples through the user’s existing and familiar voice interfaces, such as mobile phones or landlines, VAAM enables widely-accessible and cost-effective solutions for older populations over long periods of time.
There is no need to wear, learn, recharge, use their hands or buy special devices to obtain the benefits offered. Since Cardiokol’s VAAM is software-based, its scalability and affordability are unmatchable.
Does the use of voice analysis benefit the healthcare system as a whole?
Voice-based monitoring is elderly-friendly, deviceless, low-cost and scalable. It can quickly access large and inaccessible populations at risk over unlimited time periods and on a daily basis.
This can be done locally via smartphone or speaker apps, or remotely via personalized calls, call centers, and phone calls archives. It also covers hard to reach populations at risk.
Voice monitoring can also support early detection using vocal biomarkers for life-threatening health risks and can improve disease management of chronic patients.
Since oral anticoagulants can prevent approximately 70 per cent of AF-strokes and AF’s other cardioembolic complications, timely detection and management of AF patients in high-risk populations can reduce the immense healthcare socio-economic burdens associated with ischemic strokes and other complications.
Are there any teething problems or downsides to this technology?
The clinical guidelines for the diagnosis of AF requires an ECG (Electrocardiogram) test.
The VAAM is not intended to replace an ECG test as its purpose is to extend and augment limitations of current methods for AF screening and monitoring.
Once a new AF patient, or recurrent AF burden, is detected using VAAM, the patient is prompted to conduct an ECG test.
Consequently, the VAAM significantly extends the reach and the scale of AF screening, as well as improves the low detection yield of current methods, mainly in both entirely and partially asymptomatic patients.
What is the future of this technology?
The future of this technology is two-fold.
First is to screen for unknown AF and for other treatable key health risks in large, at-risk populations, primarily the over-65s by integrating VAAM into smart speakers, smartphones, digital voice assistants, connected cars, landlines, call centers, telecom clouds, and voice call archives of health providers and insurers.
Second is to develop and deploy synergistic vocal biomarkers to detect and track health conditions and risks.