Voice AI to Clinical Action: The Complete Care Pathway
AYA transforms voice into clinical intelligence. Frontline health workers use voice-based screening to capture psychological states, while constrained generative AI generates actionable clinical insights—empowering nurses, counselors, and clinicians to deliver care that reaches those who've been invisible.
Voice AI Screening: Hearing the Unheard
The foundation of care is listening. Our voice AI, trained on the Multilingual Multimodal Clinical Dataset from diverse populations across Africa, converts a 30-second voice sample into comprehensive psychological metrics—detecting indicators of depression, anxiety, and trauma that clinical staff might miss. This objective voice assessment becomes the input for constrained generative AI clinical reasoning.
60% detection rate vs. 5% current rate in primary care
Constrained GenAI: From Data to Clinical Reasoning
Voice metrics alone tell part of the story. Our constrained generative AI layer takes voice-derived psychological data and applies structured clinical reasoning—transforming numbers into actionable decisions. The system generates risk assessments, triage recommendations, and treatment pathway guidance, always constrained to evidence-based clinical protocols. 85-90% accuracy in clinical decision support, grounded in psychiatric standards and frontline worker feedback.
AI that thinks like a clinician—bounded by safety and evidence
Longitudinal Voice AI Monitoring: Sustained Healing
Recovery is a trajectory, not a moment. Repeated voice-based check-ins track psychological status over weeks and months, while constrained generative AI identifies patterns of improvement, flags deterioration, and alerts clinicians to opportunities for intervention. The system creates a living clinical record—objective voice data over time, combined with AI-guided interpretation, ensuring patients receive continuous support and supervisors gain visibility into population mental health trends.
Care that follows patients home—and keeps them connected
A Patient's Journey to Care
From a 30-second voice sample to a personalized clinical plan. Meet Amara, a woman in Ethiopia, and follow her journey through AYA's intelligent screening, analysis, and care system.
30-Voice Sample: Sharing Her Story
Amara visits her local health clinic. A community health worker conducts a voice-based screening in Amharic, asking open-ended questions about her emotional state and well-being. Over one minute, she shares her experience—sadness about recent loss, anxiety about her family's future, and difficulty sleeping.
The 30-second voice sample captures genuine emotional expression—the foundation for AI analysis.
Voice AI Analysis: Objective Psychological Metrics
AYA's voice AI, trained on the Multilingual Multimodal Clinical Dataset, analyzes Amara's voice in real-time. The system calculates a comprehensive psychiatric severity score: Emotional Distress Index (EDI) and clinical formulation
Constrained GenAI Clinical Report: Structured Clinical Reasoning
Amara's voice metrics reveal severe distress (EDI of 78). Now AYA's constrained generative AI layer activates. Using only the voice-derived data and pre-trained clinical frameworks, the system generates a detailed clinical report—not a diagnosis, but a structured guide for clinicians grounded in evidence-based protocols.
Evidence That Builds Trust
Our work isn't based on promises—it's grounded in rigorous, peer-reviewed research proving that voice AI can accurately detect mental health conditions, support trauma recovery, and save lives. These studies validate what frontline workers and patients already know: technology guided by compassion works.
An AI-enabled, trauma-informed rehabilitation protocol for Ethiopian women with complex trauma
Alemu, Y., Ohiomoba, P., Kahsay, Y., et al.
International Journal of Psychiatry Research • 2026
Key Findings
✓AI-enabled screening and rehabilitation protocol for trauma survivors
✓Effectiveness in addressing complex trauma from sex work, homelessness, and severe poverty
✓Integration with local mental health infrastructure
Implementing and analyzing the advantages of voice AI measurement-based care to address behavioral health treatment disparities among youth
Alemu, Y., Cardenas Bautista, E., Vinson, S., Ohiomoba, P., et al.Telehealth and Telemedicine Today • 2024
Key Findings
✓Voice AI measurement-based care reduces health disparities in youth mental health
✓Implementation evidence for scaling in underserved communities
✓Improved treatment outcomes through objective severity monitoring
Objectively quantifying pediatric psychiatric severity using artificial intelligence, voice recognition technology, and universal emotions: Pilot study
Alemu, Y., Teshome, S., Salegh, E., Ohiomoba, P., & Vinson, S.
Annals of Research Protocols • 2023
Key Findings
✓Voice AI accurately quantifies psychiatric severity in pediatric populations
✓Universal emotional recognition improves diagnostic accuracy
✓Pilot validation across diverse populations
Objectively quantifying pediatric psychiatric severity using AI and voice recognition technology
Caulley, D., Alemu, Y., Burson, S., et al.
JMIR Research Protocols • 2023
Key Findings
✓Validated AI and voice recognition methodology for psychiatric assessment
✓High accuracy in identifying clinical severity levels
✓Scalable approach for global mental health screening