Prentice Tom, CMO at Kintsugi - the AI-backed startup that can detect signs of depression and anxiety just by listening to short clips of speech

A new report by Mental Health America (MHA) states that over 50 million Americans are currently experiencing mental health conditions. These individuals span all walks of life, from young people dealing with climate anxiety or the adverse effects of social media to the impact of chronic disease and the loneliness epidemic’s toll on the elderly. 

The worst part is that most of these people suffer in silence. Stigma and fear of judgment still sadly prevent many from seeking help. When patients don't speak up, it's difficult for their healthcare providers to know they are struggling. Primary care physicians correctly detect mental health conditions just 47% of the time and are likely to note them down only 33% of the time. As a result, mental illnesses are one of the most underdiagnosed conditions in medicine, and an estimated 60% of people who suffer from mental health conditions never receive treatment.

Screening for mental health conditions needs to be included as part of every primary care visit. In fact, the US preventive task force recently mandated that all patients under 65 should receive depression and anxiety screenings. In addition, hospitals routinely monitor physical conditions, like blood pressure, weight, and temperature, so why not a patient's mental health? 

The reality is that mental health evaluation has traditionally been as much art as science. It is an area with few objective and quantitative tools available to the clinician to assess the presence of disease. 

Existing screening processes are time-consuming, cumbersome, and rely on paper-based forms that can take up to 15 minutes per patient to complete. As a result, depression screenings take place in less than 5% of primary care interactions. 

Mental health conditions also tend to lack measurable indicators of disease, known as biomarkers. For example, paper-based tests screen for depression and anxiety based on the patient's self-reported experience. However, true biomarkers are more objective and quantifiable. One of the best-known examples is a high blood glucose level, which helps clinicians detect diabetes and determine its severity.

Despite the very real symptoms that patients experience, the medical field tends to ignore health conditions that are difficult to measure, like long covid or chronic fatigue syndrome. As a result, it can take many people suffering from these conditions before they are officially recognized. For individuals, it can take years and numerous visits to healthcare professionals  before they receive the correct diagnosis. 

‍Introducing vocal biomarker technology 

A mental health diagnostic tool that is objective and easy to use has been lacking until now; this is where voice biomarkers come in. Using voice biomarkers as a diagnostic tool is not a recent discovery. Scientific papers dating back to the 1970s describe the subtle vocal indicators of these health conditions, including reduced range of pitch and volume and more pauses in speech. 

Advances in computational power and artificial intelligence have enabled us to turn this academic knowledge into a diagnostic tool that can seamlessly integrate into clinical workflows. For example, while seeing their doctor, patients can simply speak into their phone or recording device. Kintsugi’s voice biomarker technology can determine how depressed or anxious a patient is in real-time, helping clinicians ensure their patients receive the proper support in their moment of need. 

In addition, voice biomarkers are extremely accurate at detecting the presence of disease – much more so than just talking to people. Even when people are trying to hide it, or it's buried in their emotional state at that time, we can still pull out the psychiatric pathology from certain voice characteristics.

The implications of this tech are huge: not only for healthcare systems, which can use it to gain a 360-degree view of a population's mental health but also for individuals, who can better assess their mental wellness and take action to promote it. In both cases, users are alerted about lapses in mental health no matter how “mild” they may seem. These early stages of mental distress are when we should be taking action. 

Kintsugi is bringing diagnostic precision to mental health, helping to elevate it to the same status as our physical health. The result is a scalable, quantifiable, reproducible, and non-invasive tool that can screen every individual in the US for mental health conditions. 

‍Use cases

In addition to embedding into routine physical wellness exams, voice biomarker technology can assist in telehealth and call centers. Telehealth is increasingly becoming a vital healthcare touchpoint for the most vulnerable populations – such as people with disabilities or those living in rural areas – who cannot easily and regularly visit a primary care physician. Our platform runs seamlessly in the background with the caller's permission and can let providers know whether these patients might benefit from mental health services.

The COVID-19 pandemic has made virtual health visits more popular than ever. But it can be even more difficult for these practitioners to detect signs of mental illness. We need to ensure that patients get the same level of care as they would during traditional face-to-face visits. With Kintsugi, clinicians can screen patients for mental health conditions without interrupting the session. Our algorithm analyzes the patient's voice as they speak (not what they are saying, but how they say it) and provides a mental wellness score in real-time. 

Imagine a patient arranges a virtual visit for ongoing migraines or back pain. It has been known for years that chronic pain has high comorbidity with depression, but it is not always easy for the physician to diagnose, especially in a virtual environment. Tools like Kintsugi can help clinicians identify that the patient’s pain might be impacting their mental health, ultimately painting a fuller picture of the patient for proper treatment. 

Our API can be integrated into existing clinical workflows, EHRs, call-centers, telehealth platforms, and remote patient monitoring apps, supporting home care beyond the traditional healthcare setting. Facilitating screening, triage, and scheduling is one way to streamline and operationalize access to mental healthcare across the health system.

‍A paradigm shift in mental health

The Kintsugi tool has the potential to change how we view mental health. Traditionally, our society's approach has been physician-centered and tends to focus on disease. When we can assess everyone's mental health, we can broaden the scope of what's possible. Rather than just focusing on one end of the mental health spectrum – or, more specifically, the absence of health – we can stratify the population based on mental wellness. 

We can also stratify care options. For example, many patients don't require pharmacologic therapy but could benefit from a meditation app, journaling, counseling or perhaps even an exercise regimen. In addition, if people can reliably assess their mental wellness regularly, they are more empowered to take action. Just like counting your steps provides quantitative feedback, using voice biomarker technology to measure your mental health and wellness can allow you to make minor adjustments to ensure that your health stays on track.

Finally, we all experience moments of feeling low. Tools to track mental health can help people better identify when "normal" sadness has transitioned into clinical disease. In addition, monitoring symptoms is significant for first-time sufferers, such as postpartum moms or those recently diagnosed with a chronic condition, who may believe that what they're experiencing and feeling is something all people go through. It's time to give these people a voice.