Heidi Christensen

Heidi Christensen

Professor at the University of Sheffield, UK

CcHAT: CognoSpeak: a Cognitive Health Assessment Tool

CcHAT (CognoSpeak: a Cognitive Health Assessment Tool) was a major clinical research initiative funded by the NIHR i4i (Invention for Innovation) program. At its core, the project addressed a critical bottleneck in the UK healthcare system: the overwhelming pressure on Memory Assessment Pathways (MAPs). By utilising natural language processing and machine learning, we developed a system capable of performing automated, remote cognitive screenings—saving clinician time and accelerating the diagnostic journey for patients.

During the course of the project, our branding evolved to reflect the expanding scope of our research. While CognoSpeak™ now serves as the overarching name for our entire suite of speech-based assessment projects, the specific tool developed and validated for the dementia detection pathway within this project is now known as CognoMemory.

Team and Interdisciplinary Collaboration

The success of CcHAT and the launch of CognoMemory were built on a deeply integrated partnership across academia, the NHS, and the commercial sector:

  • Clinical & Academic Leadership: Led by Dr. Dan Blackburn (University of Sheffield/STH) and Prof. Heidi Christensen (University of Sheffield).
  • Commercial Development: Therapy Box, who transformed our research models into a robust, user-facing digital platform.
  • Clinical Translation: Devices 4 Dignity (D4D), providing strategic guidance on clinical safety and NHS adoption.
  • Diverse Recruitment Networks: We collaborated with 10 NHS Recruitment Sites (including CNWL, Newcastle, and Manchester) and four key community organisations, Israac Somali Community, ShipShape, Sheffield Chinese Community Centre, and Meri Yaadein to ensure our AI was trained on a truly representative dataset.

Key Outcomes and Successes

The project concluded in July 2025, having fundamentally shifted the landscape for speech-based biomarkers:

  • Representative AI: We achieved 100% of our recruitment targets for minority ethnic groups, addressing a major gap in AI fairness. Our system now shows no significant performance drop-off across different accents and backgrounds.
  • Clinical Benchmarking: Validated against gold-standard assessments (MoCA, MMSE, and ACE-III), demonstrating that CognoMemory can reliably predict clinical scores.
  • Media & Recognition: Featured extensively by the BBC and awarded for excellence in inclusive research design.
  • Scaling Up: The project successfully demonstrated the feasibility of using the underlying CognoSpeak technology for broader conditions, leading to the launch of CognoStroke and CognoMND.

Key Publications

The academic impact of CcHAT is reflected in several flagship 2025 publications:

  • Pahar, M., et al. (2025). “CognoSpeak: an automatic, remote assessment of early cognitive decline in real-world conversational speech.” IEEE Journal of Biomedical and Health Informatics.
  • Tao, F., et al. (2025). “Early dementia detection using multiple spontaneous speech prompts: The PROCESS challenge.” Proc. ICASSP 2025.
  • Pan, Y., et al. (2025). “A two-step attention-based feature combination cross-attention system for speech-based dementia detection.” IEEE/ACM Transactions on Audio, Speech, and Language Processing.
  • Mirheidari, B., et al. (2025). “Automatic Detection and Sub-typing of Primary Progressive Aphasia from Speech.” Proc. Interspeech 2025.