Resource · Wearables

How wearable data helps in behavioral health.

Heart rate variability, resting heart rate, sleep stages, activity, and skin temperature are now collected continuously by tens of millions of clients on Fitbit, Oura, Whoop, and Google Health Connect. For behavioral health, this is the first time the body's signal has been available between sessions. Here's how Vybz uses it to detect stress, depression, and risk earlier, with research-backed models, LLM reasoning, and clinician-governed intervention.

Research

Models + LLMs

How we access data

Research

Stress and depression leave a signal on the body.

Our peer-reviewed behavioral health research shows that consumer wearables capture enough physiological signal to detect stress and signs of depression, before the client can put it into words.

In two peer-reviewed studies, we showed that wearable signals are enough to flag stress and depression risk — heart rate alone detected stress with 73% accuracy, and heart rate variability detected stress (a key indicator of depression) with 96.2% accuracy. No ECG, no clinic visit, no patient self-report required. The signal is in the body; the question is whether anyone is listening between sessions.

Read our work on Google Scholar

Findings from our research

Wearable signals detect mental health risk

Stress

73%

Accuracy

Heart rate alone is enough — no ECG, no HRV — turning any consumer wearable into a passive stress sensor.

Depression

96.2%

Accuracy

Heart rate variability surfaces the earliest physiological signature of depression, before symptoms cross the clinical threshold.

Models + LLMs

LLMs reason. Our models detect the signal.

Large language models are good at reasoning over structured data, like sleep, activity, and physiological trends. They are not good, out of the box, at predicting risk around specific mental health symptoms.

Vybz Health uses proprietary behavioral health models, trained on clinical cohorts, that look for the specific physiological and behavioral signals associated with stress, depression, and other mental health risks. Once a signal is detected, the structured result is passed to an LLM that reasons over it, decides what context matters, and orchestrates our AI agents to follow up with the client or escalate to the clinician.

See how AI intervention works

Raw wearable signals

HRV, resting HR, sleep stages, activity, circadian.

Vybz proprietary models

Stress + depression risk classifiers, tuned on clinical cohorts.

LLM reasoning + orchestration

Routes the signal into agent action and clinician escalation.

How we access wearable data

No patient app required. The client just grants access.

Vybz does not require a clinic-branded patient app or any new device. The client gives the Vybz agent the read-only permissions it needs to pull data from Fitbit, Oura, Whoop, and Google Health.

Once permission is granted, our agents continuously analyze the wearable data to proactively detect the signals that matter for behavioral health — rising stress, sleep disruption, activity decline, circadian shift — and reach out at the right moment. The client never downloads anything. The clinician sees only the structured signals, not raw biometric streams.

Supported wearables

Plug in once. Read-only access.

Fitbit

Oura

Whoop

Google Health

How wearable data helps

The body's signal, finally working for behavioral health.

Continuous physiological signal

Heart rate variability, sleep stages, daily activity, and circadian patterns from the client's existing wearable, sampled around the clock, not just in the room.

Research-backed risk detection

Our peer-reviewed behavioral health work shows wearable signals can flag stress and depression risk earlier than self-report. Models are tuned on clinical, not consumer, cohorts.

LLMs orchestrate the response

Structured signals from our proprietary models are passed to the LLM, which decides which agent reaches out, which question to ask, and when to escalate to the clinician.

See it for yourself

Book a demo to see wearable signals in a Vybz encounter report.

Walk through how Vybz pulls wearable data, applies our proprietary behavioral health models, and triggers the right intervention at the right time.

Book a demo