top of page

How Ellipsis Health’s clinically validated vocal biomarker technology is breaking down barriers in the mental health sphere

Ellipsis Health leads the field in vocal biomarker technology due to our pioneering research into natural language processing (NLP) and acoustic models that span a range of age groups and cultural backgrounds. We have worked diligently to create a product that identifies depression and anxiety symptom severity early and facilitates monitoring over time ultimately leading to more positive patient outcomes at a lower cost.

Without our industry-leading research, we could never have become the only clinically validated vital sign for mental health. But it’s the applications of vocal biomarkers in the mental health space that really make a difference to patients and healthcare organizations.

Using the analytical prowess of artificial intelligence, and drawing on vast databases of past examples, we have redefined how we detect, monitor, and treat mental health issues. Here, we’re going to go into more detail on these breakthroughs, exploring what they mean for the patient and how they improve care.

Early detection of mental health disorders

The adage “an ounce of prevention is better than a pound of cure” is especially significant when it comes to mental health issues. Psychiatric problems that go unreported or are diagnosed late tend to snowball, making treatment at a later stage much less efficient and much more costly.

According to a study by The Lancet, around half the world’s population will suffer from some form of mental disorder during their lifetime. Usually appearing for the first time in childhood, adolescence, or young adulthood, it is critical to address these issues early to prevent escalation into problems that can affect a patient’s personal and professional life, as well as their physical health.

Ellipsis Health’s patented assessments are accurate and simple to use, quickly alerting individuals or healthcare providers to potential disorders. Available through health plans and health systems embedded in case management and telemedicine calls, the focus on easy accessibility makes AI-powered vocal biomarker technology a more scalable early detection method than self-reported paper-based surveys and lengthy provider assessments.

Consistent, accurate mental health monitoring

Continuous monitoring is vital in managing mental health effectively and intervening when problems arise. However, traditional mental health checkups are insufficient in this regard as they typically occur at intervals that could have gaps of many months between them. This periodic approach only captures a snapshot at a specific time, providing an incomplete picture of a patient’s mental state.

Applying vocal biomarker technology in regular mental health checkups allows healthcare professionals to more frequently monitor mental health indicators through a 60-second, non-invasive voice recording. This voice technology can be integrated whenever a user is speaking; embedding in apps or digital front doors, and also layered onto phone calls. Rather than relying on self-reported surveys, we screen patients using the semantic signals in speech (what is said) and the acoustics (how it is said). Easy and more time efficient for the patient, the regular monitoring process helps to pick up on all the subtleties and fluctuations inherent in mental health conditions.

With a more comprehensive stream of data, our machine-learning technology can understand the baseline for each individual and single out anomalies. This is a crucial tool in anticipating potential flare-ups or downturns in a patient's mental health. For instance, a change in speech patterns could signal an impending depressive episode or heightened anxiety, allowing healthcare professionals to intervene proactively.

Increased access for under-represented communities

A 2018 study showed that while almost half of white people suffering from a behavioral disorder received mental health treatment, the same was true for just 31% of Black and Hispanic people and 22% of Asians. This is explained in part by systemic inequalities, such as the fact that over 50% of counties in the U.S. lack a single psychiatrist, with rural areas especially under-served.

However, even when there is a mental health provider nearby, assessment can prove difficult if the patient is from a demographic minority. To address this, Ellipsis Health has built uniquely large databases containing tens of thousands of recordings from people of diverse backgrounds. Our peer-reviewed study in IEEE Xplore backs the resulting algorithm as a “viable technique” and shows that it is generally accepted by patients of different ages and races.

Available at home and fully confidential, vocal biomarker technology provides unprecedented access to mental health screening, breaking down geographical and cultural barriers. Where time and financial constraints prevent patients in under-served areas from seeking mental health attention, telehealth consultations, case management calls, and mobile apps can significantly increase access. Similarly, algorithms assist mental health professionals in triaging patients accurately and objectively, regardless of their background.

Improved communication through speech recognition

Beyond the specifics of anxiety and depression detection, the applications of vocal biomarkers stretch to patient-provider communication as a whole. Designed to work as a clinical decision support tool in a healthcare provider’s arsenal, Ellipsis Health’s technology can do a lot of the administrative legwork by screening patients before the visit. This way, mental health professionals[4]  – including therapists, psychiatrists, care navigators, case workers, and more – can use their valuable patient time more effectively.

But it’s not just an advantage from a practical point of view. With a fuller picture of their patient’s emotional and mental state, providers[5]  can be more understanding and empathetic in how they deliver care.

Enhancing traditional mental health assessments

Technologies like Ellipsis Health’s voice analysis not only support the human side of mental health care providers, but they also provide additional data to frameworks like PHQ-8 and GAD-7 surveys. These questionnaires are susceptible to inconsistencies, relying on self-reporting to identify issues that many people suppress or at times exaggerate.

One drawback of traditional assessments is that patients often know what response they are “supposed” to give and steer the survey if they feel a stigma attached to the way they are feeling.  Ellipsis Health uses a conversational approach, where patients choose from a range of day-to-day subjects and speak freely for around 60 seconds into a device. The technology can also be layered seamlessly onto calls between a provider and a patient, monitoring ambient conversation. Analyzing both syntax and vocal patterns, we offer a more objective, engaging, and data-based evaluation than traditional assessments. This allows healthcare providers at times to go deeper than what patients are prepared to reveal openly.

In practice, if a patient’s self-reported data in a PHQ-8 survey is inconsistent with their vocal biomarker analysis, it might signal the need for further exploration or alternative assessment approaches.

AI-enabled technology that increases the efficiency of healthcare systems

Rather than adding another complication to already convoluted workflows, at Ellipsis Health, we work hard to remove barriers to treatment. The main ways are through seamlessly integrating our digital front door application and call analysis technology within health systems’ existing workflows.

We are validating our machine learning-based smartphone apps for children and adolescents, as well as expanding their use in senior populations, perinatal, and other specialty areas. Through push notifications, effective onboarding, and a simple user experience, we have increased the time users spend engaging with the technology allowing for better monitoring of mental health conditions.

Ellipsis Health also collaborates with healthcare providers and academic medical centers, providing them with mental health screening technology. For example, we have partnered with Ceras Health, a digital health solutions provider, to screen chronically ill patients for depression. With just a few minutes of speech, Ceras Health receives a score that represents a patient’s level of depression, as well as a more detailed analysis that is mapped onto traditional PHQ-8  survey measurements with which providers are familiar.

These applications of vocal biomarkers have already gone a long way to changing mental health care as we know it. Even so, at Ellipsis Health, we know we’re only at the beginning of this exciting journey.


[1]  Zhang, A.; Kamat, A.; Acquati, C.; Aratow, M.; Kim, J.S.; DuVall, A.S.; Walling, E.“Evaluating the Feasibility and Acceptability of an Artificial-Intelligence-Enabled and Speech-Based Distress Screening Mobile App for Adolescents and Young Adults Diagnosed with Cancer: A Study Protocol.” Cancers 2022, 14, 914.

[2] Lin, D; Nazreen, T; Rutowski, T; Lu, Y; Harati, A; Shriberg, E; Chlebek, P; and Aratow, M. “Feasibility of a Machine Learning-Based Smartphone Application in Detecting Depression and Anxiety in a Generally Senior Population,” (2022) Front. Psychol. 13:811517. doi: 10.3389/fpsyg.2022.811517

[3] Lin, D; Nazreen, T; Rutowski, T; Lu, Y; Harati, A; Shriberg, E; Chlebek, P; and Aratow, M. “Feasibility of a Machine Learning-Based Smartphone Application in Detecting Depression and Anxiety in a Generally Senior Population,” (2022) Front. Psychol. 13:811517. doi: 10.3389/fpsyg.2022.811517


bottom of page