Naver Cloud AI screens for dementia using speech and content analysis

In May 2026, Naver Cloud showcased its “Naver Care Call” software at the Local Government and Public Week in Tokyo, Japan, utilizing speech patterns and conversational content to screen for early signs of dementia and other cognitive conditions.

AI Analyzes Speech Cues to Flag Cognitive Decline

  • Voice Biomarkers: Artificial intelligence software analyzes non-verbal speech cues—such as word-recall pauses and speaking speed—that traditional transcription tools normally ignore as background noise.
  • Data Integration: Combining speech-feature extraction tools with large language models helps distinguish healthy individuals from patients with cognitive decline by tracking vocabulary, sentence structure, and narrative coherence.
  • Clinical Validation: Ongoing trials involving hospital partnerships aim to determine if automated phone interactions can accurately flag early cognitive decline without requiring in-person cognitive testing.

Interspeech 2026 Presentations Highlight Speech Feature Extraction

Researchers from Naver Cloud presented two distinct papers at the Interspeech 2026 conference in Australia, detailing methods to differentiate dementia patients from healthy cohorts by evaluating both verbal content and speaking delivery. The technical approach emphasizes examining not just the literal vocabulary used during a conversation, but the behavioral mannerisms of speech production.

Standard speech-to-text transcription software frequently strips away hesitation markers, word-finding pauses, and conversational pacing, treating these features as audio artifacts. However, the Naver Cloud research team used OpenAI’s Whisper model to extract raw acoustic speech features. They subsequently analyzed vocabulary choice, sentence structure, and narrative coherence using the large language model GPT-5.2.

When researchers evaluated datasets in isolation, transcripts alone yielded lower diagnostic performance. In a secondary study, predictive accuracy increased from 81.48 percent using text transcripts alone to 90.14 percent after sequentially incorporating narrative themes, speech speed, pauses, and pronunciation data.

Clinical Trials at Seoul Boramae Hospital

To move beyond theoretical modeling, Naver Cloud is actively conducting clinical research alongside Emocog and Seoul Boramae Hospital. These trials evaluate whether automated telephone interactions can reliably capture early cognitive decline from conversational data alone.

The clinical testing builds directly upon the public demonstration of “Naver Care Call,” which was featured at the Local Government and Public Week 2026 exhibition in Tokyo during May.

Global Market Projections for Voice Biomarkers

The integration of voice biomarker technology into routine medical workflows is expanding beyond experimental research settings. According to market research projections cited by The Business Research Company, the global voice biomarker market is expected to grow from 3.35 billion dollars in 2025 to 7.41 billion dollars by 2030.

References

  • Naver Cloud. Research papers presented at Interspeech 2026, Australia.
  • The Business Research Company. Global Voice Biomarker Market Report, 2025–2030.
Photo of author

Priya Deshmukh - Senior Editor, Health

Priya Deshmukh Senior Editor, Health Deshmukh is a practicing physician and renowned medical journalist, honored for her investigative reporting on public health. She is dedicated to delivering accurate, evidence-based coverage on health, wellness, and medical innovations.

Yan Diomande must reclaim form for Real Madrid, Mundo Deportivo says