OSP: Could you please share the ‘elevator presentation’ description of Vocalis—who you are, what you do, and what sets you apart?
SH: Vocalis Health is an AI health tech company pioneering the development of vocal biomarkers – where health-related information is derived from analysis of people’s voice recordings – to screen, detect, monitor and predict health symptoms, conditions and diseases.
In 2016, I co-founded Vocalis Health with Daniel Aronovich, chief technical officer of the company. While treating patients in the hospital, I realized that I was habitually listening to their voices to gauge the state of their health.
As there was no standard system of quantifying their disease based on the voice, I used my subjective experience to measure their health status. I then realized that I could help millions of patients before they even entered the hospital by utilizing artificial intelligence and vocal analysis technology to screen patients for a variety of diseases.
The human vocal system includes the lungs and the lower airways that provide air supply. The vocal folds modulate the airflow through vibration, which produces the voice source; this sound is then modified by the tongue and mouth.
When thinking about vocal analysis and the conditions we may be able to detect, monitor or screen for, the most obvious conditions are those that impact the respiratory system as they affect these organs and thus, can be detected through the voice. We created Vocalis Health with the mission of standardizing voice analysis to raise the alerts as early as possible to improve patient outcomes, and to do so in an accessible, cost-effective, and non-invasive way.
We are currently focused on screening users for COVID-19 and pulmonary hypertension (PH) and on monitoring patients with chronic diseases such as chronic obstructive pulmonary disease (COPD) and congestive heart failure (CHF). We believe that our vocal biomarkers can be applicable to a myriad of diseases, ranging from acute to chronic, and allow doctors to be proactive in their treatment of patients.
OSP: Could you please explain what a vocal biomarker is, how data has been applied to date, and how you expect vocal biomarkers might be put to use in the future?
SH: Vocal biomarkers are signals that can be detected in the voice. When we record a person’s voice, we then transfer that recording into the visual domain, into an image called a spectrogram. We then utilize machine learning algorithms to assess the correlation between a patient’s voice/vocal patterns to a variety of diseases, symptoms and medical conditions, such as shortness of breath related to a COPD exacerbation, or the presence of Pulmonary Hypertension. This analysis is performed using computer vision techniques to find small changes in the spectrogram.
Depending on the biomarker, our platform can be used to screen, detect, or monitor health conditions or symptoms. For COVID-19, we have created a COVID-19 “footprint;” when we screen a person for COVID, we compare their voice to the COVID-19 composite image, which is based on thousands of COVID-19-positive vocal data points, and based on the correlation, we can determine with over 80% accuracy, the person’s risk of being COVID-19 positive. This layer of screening is crucial for better allocated diagnostic and health personnel resources across health systems.
In terms of future applications, we are exploring the use of vocal biomarkers for monitoring congestive heart failure, pulmonary hypertension, depression, sleep apnea, asthma, and more. In all of these cases, the process is the same for developing the vocal biomarkers specific to each condition.
The AI algorithms are developed using the voice recordings of people diagnosed with a specific disease or condition. Their voice recordings, along with other medical information, are collected in clinical trials, are then transformed to spectrograms, and classified using machine learning to “train” a vocal biomarker.
The biomarker is then tested with a different group of patients to see how well it performs, measuring how well the insights it provides match the clinical information about this “test” group. Over time, Vocalis Health collects more and more data, which strengthens the AI algorithms and makes the vocal biomarker even more accurate.
OSP: Please share your perspective on COVID-19 screening tests and technology out there, and how VocalisCheck improves upon options available to test for the virus.
SH: Currently, many health systems and governments are screening large populations using PCR tests and rapid antigen tests. These types of tests fall into the more costly “diagnostic” category. They are resources heavy, more time-consuming, and require medical personnel to perform the test and analysis. Using diagnostic approaches for screening large populations is not a sustainable model.
On the other end of the spectrum are symptom checking approaches. Asking a person if they have had a fever, a cough or other symptoms. This type of screening approach is flawed as well.
Firstly, those who want to circumvent their symptoms (i.e., to go to work) can take over-the-counter medications like Tylenol to mask their symptoms. Secondly, these symptoms can be shared with many other illnesses. Hence, their accuracy is low in detecting COVID in patients.
onversely, our COVID-19 vocal biomarker has now proven to perform with over 80% accuracy in identifying COVID-19 risk in both symptomatic and asymptomatic patients. This voice-based screening can be done on any connected device (smartphone, tablet, computer), can be performed at home or elsewhere at no additional cost, and with no extra equipment aside from a connected device. This makes the screening solution highly scalable and cost-effective, and it can enable health systems and governments to better allocate diagnostic resources to those identified as having the highest risk of COVID-19 infection.
OSP: How does the reliability of this screening tool compare to other currently available COVID diagnostics?
SH: VocalisCheck is not a diagnostic tool, but rather a highly scalable screening tool that can fill a significant gap in the current approach to COVID screening, with the ability to effectively funnel those with a high risk of infection to the appropriate diagnostic test. We developed our COVID-19 vocal biomarker as a highly scalable and non-invasive screening tool that can help extend the reach and reliability of screening and better inform and guide diagnostic testing.
Vocal biomarker testing is not intended to provide a definitive diagnosis, but to help health systems triage potential cases by rapidly identifying people who have a higher probability of being COVID positive and require more advanced diagnostic testing, quarantine, or in-person medical care. This totally non-invasive screening tool helps physicians remotely assess the health of their patients, providing immediate results which can guide clinicians in making fast, educated decisions about diagnosis and treatment plans.
OSP: What’s next for VocalisCheck—do you have any arrangements to use it in the field, in hospitals or clinics?
SH: In addition to supporting hospitals and health systems to better funnel diagnostic resources with more accurate screening, VocalisCheck is being utilized by businesses looking to safely return their workforces to in-person operation. Employees simply login to the app in the morning to give a voice sample.
The recording is sent to the cloud, analyzed, and within a minute, the person is given a COVID risk score. If the person’s risk score is low, they may safely go to work. However, with a high risk score, that person may require further medical attention.
We see this as a key application for VocalisCheck – to help businesses, governments, universities and society at large safely “return to normal” as it is likely that we will be living alongside this pandemic for another few years.
OSP: Is there anything else you’d like to add we didn’t touch upon above?
SH: COVID-19 is an excellent application for our technology. However, vocal biomarker technologies have only scratched the surface at this point. I outlined some potential use cases above – COPD, CHF, pulmonary hypertension, among others. Voice-based remote patient monitoring may have a significant impact on healthcare, supporting virtual/telehealth, which is on the rise, and serving as an adjunct health technology to monitor patients in clinical trials for drug development and beyond.