Models

thymia offers two voice biomarker models, each designed for a distinct use case. Both analyse neural morphemes® extracted from natural speech and return clinically validated scores. They differ in their intended setting, output type, regulatory status, and recording requirements.

Model overviews

Helios

Wellbeing and safety monitoring

Wellbeing monitoring from live voice. Detects distress, stress, fatigue, low self-esteem, and burnout in real time — across human and voice agent interactions — for healthcare, pharma, and safety-critical industries.

Apollo Clinical

Clinical-grade health detection

Voice-based depression and anxiety assessment covering all DSM-V criteria with disorder-level results and granular symptom scores. Apollo is a regulated medical device intended for qualified healthcare professionals only.

Comparison

Which model should I use?

Use Helios if you are:

• Monitoring operator wellbeing or fatigue in safety-critical environments
• Tracking patient wellness between clinical appointments
• Embedding continuous mental wellness monitoring into a consumer or enterprise product
• Running longitudinal wellbeing programmes in pharmaceutical or occupational health contexts

Use Apollo if you are:

• A healthcare professional conducting remote mental health triage
• Screening patients for depression or anxiety before a clinical consultation
• Tracking symptom-level treatment response over time
• Building a regulated clinical decision support tool

Using models via Sentinel

Both Helios and Apollo are also available as real-time streaming biomarkers via Sentinel — thymia’s monitoring layer for live voice AI applications. Rather than submitting recordings via the REST API, Sentinel streams audio from your existing voice stack and returns biomarker signals in real time during the conversation.