Vocal biomarker use can elevate patient monitoring: Sonde Health

Vocal biomarkers can be highly useful, telling caregivers and researchers things that other vital signs are not able to fully show. Outsourcing-Pharma recently spoke with David Liu, CEO of health technology company Sonde Health, about the potential of vocal biomarkers in research and care, and how the company’s voice-enabled symptom detection and monitoring platform can be useful in watching chronic and mental health conditions.

OSP: Could you please share a bit about Sonde Health—who you are, what you do, key capabilities, and what sets you apart from other companies working in this space?​

DL: Leveraging 1m+ voice samples, Sonde Health’s proprietary voice-based technology is designed to detect changes in health conditions – like mental fitness and respiratory disease – from changes in voice. Using advanced audio signal processing and machine learning, Sonde senses and analyzes subtle vocal changes due to a change in a person’s physiology to provide early health detection and monitoring.

What differentiates Sonde Health from other companies working in this space are:

• Data: The quality, breadth, and size of our voice data set are unparalleled, enabling us to create models that are accurate and unbiased. In addition, the majority of our voice data is health-labeled by clinicians.
• Audio signal processing and voice expertise: We analyze spectrographs, extract acoustic features, use speech science and pathology, and create machine learning models to produce health insights.
• Research-based science: Our vocal biomarker-powered detection and monitoring APIs for multiple health conditions (respiratory and mental health) have been developed using primary and secondary research.
• Commercial applicability: Sonde has partnerships with wearable OEMs, telehealth, medical device, automotive, pharma companies, and payors to explore a variety of vocal biomarker use cases.
• Embedded: Sonde is the only vocal biomarker company that has technology embedded in Qualcomm smartphones and wearable chipsets so voice can be analyzed on the device and in the background.

OSP: Please tell us about biomarkers, i.e. what they are, how they have been used in clinical research and patient monitoring, etc.​

DL: A biomarker is a measurable physical indicator of a person’s health. Common biomarkers come from your heart rate, blood pressure, mucus, and blood. In research settings, biomarkers are often used to measure the efficacy of medical treatments. In clinical practice, medical providers can use biomarkers to understand and monitor the severity of their patient's condition and whether that condition is improving.

With the advent of wearable technology, anyone can track and monitor their health in real-time through digital biomarkers. These are biomarkers that can be captured, stored, and analyzed by a digital device, such as blood oxygen level from a wearable or eye-tracking from a digital camera.

Vocal biomarkers are a growing and significant portion of the digital biomarker space. These signals, which come from the sound of someone’s voice, can be captured and analyzed through digital devices. Rich signals and data can come from vocal biomarkers, which can provide useful health insight as well as be used in predictive modeling for detection and monitoring.

OSP: Why might vocal biomarkers be of particular use in relation to Alzheimer’s and other conditions involving cognitive dysfunction?​

DL: 12-18% of adults over the age of 60 are impacted by mild cognitive impairment (MCI), and 10-20% of those with MCI develop dementia-related diseases like Alzheimer’s. Vocal biomarker technology makes early detection of MCI possible by detecting subtle and even inaudible variations in the human voice. With this knowledge, patients and their care providers can act earlier to manage their condition and take steps to mitigate disease progression.

OSP: Please share how vocal biomarker data typically is collected in patients and clinical trials, and how monitoring/data collection technology has evolved in recent years.​

DL: Not all vocal biomarker data is equal because they are not collected in the same way – different companies use different methodologies. Sonde has and continues to collect voice samples in clinical/research settings where clinicians provide the health labels.

If voice samples are not health labeled, the data is not reliable, and neither are the models that they power. From health-labeled data and machine learning modeling, Sonde can stratify and engage patient populations to find the best outcomes for individuals.

OSP: You mention vocal biomarkers might be of use to industries beyond life science—could you please tell us a bit about that?​

DL: Vocal biomarkers can be integrated into smartphones/wearables for wellness applications. For example, Sonde’s Mental Fitness capability can be used by wellness companies to provide smartphone and wearable users with voice-based wellness data in new and pre-existing applications.

In the automotive and transportation industries, vocal biomarkers can be used to detect driver impairment due to intoxication, marijuana, sleep deprivation, etc. This is particularly relevant given Joe Biden’s mandate requiring auto manufacturers to install anti-drunk technology in new cars as early as 2026.

In healthcare, vocal biomarker technology can be embedded into wearable devices like hearing aids to detect signs of MCI. This is something Sonde is currently exploring with GN Group.

OSP: Where do you see vocal biomarker technology and industry use of vocal biomarkers headed in the future?​

DL: People have primarily remained reactive about their healthcare, as opposed to proactive. Our goal is to integrate vocal biomarker technology into wearable devices, smartphones, and other devices so that people will have even greater access to actionable health insights that allow them to take a proactive approach to their healthcare.

What is unique about vocal biomarker technology is that it does not require an additional device. As a result, anyone with a smartphone or hearing aids, for example, has access to voice-based health monitoring. Combined with the improved telehealth tools that have emerged due to the COVID-19 pandemic, medical providers can get vital insights into patients in even the most remote areas, thereby expanding access to preventative healthcare across the globe.