Research & evidence
Built on research. Developed for real-world impact.
DementiaDetect takes a research-led approach to multimodal AI, digital biomarkers and earlier cognitive risk detection. We believe credible claims must be earned through evidence, and developed transparently.
Our research focus
How everyday signals may relate to cognitive change
DementiaDetect explores how patterns across speech, sleep, activity, cognition and behaviour may help identify earlier signs of cognitive decline.
The work draws on health informatics, machine learning and the growing field of digital biomarkers. The emphasis throughout is on responsible, explainable methods and on understanding the limits of what the data can and cannot tell us.
Dementia and cognitive decline
Multimodal digital biomarkers
AI and machine learning
Ethical data use
Validation roadmap
Developed through staged validation
No single study makes a clinical tool trustworthy. The platform is intended to be developed step by step, with claims kept proportionate at every stage.
Technical validation
Establishing model performance, reliability and limits before clinical claims.
Retrospective testing
Evaluating against existing consented datasets to understand real-world signal.
Prospective studies
Designing forward-looking evaluation with academic and clinical partners.
Clinical pilots
Working within real care pathways to assess workflow fit, safety and benefit.
Real-world evaluation
Measuring impact and value over time, and publishing as the evidence base grows.
We welcome collaboration with
- Universities and research groups
- NHS organisations and memory services
- Clinicians and dementia researchers
- Data scientists and AI researchers
- Patient and public involvement groups
Collaboration
Better research happens together
DementiaDetect welcomes partnerships with universities, NHS organisations, clinicians, dementia researchers, data scientists and patient groups.
We are interested in validation studies, data partnerships, publication and clinical evaluation, all conducted with appropriate ethics and governance.
Statements on this page describe research intentions and a development roadmap. They are not claims of validated clinical performance. We will share evidence as it is generated, and keep all claims proportionate to validation.
Evidence roadmap
How we intend to earn clinical trust
A transparent map of the questions we need to answer, the evidence we intend to generate, and where each stage stands today.
| Stage | Question | Evidence to generate | Status |
|---|---|---|---|
| Technical validation | Does the model perform reliably? | Accuracy, sensitivity, specificity and bias testing | In progress |
| Retrospective testing | Can signals identify known patterns? | Consented dataset testing and model benchmarking | Planned |
| Prospective study | Does it work with real users over time? | Longitudinal, real-world evaluation | Seeking partners |
| Clinical pilot | Does it support workflow and decisions? | Clinician feedback and pathway impact | Seeking partners |
| Real-world evaluation | Does it improve outcomes or efficiency? | Health-economic and safety data | Future |
Status reflects our current development intentions and is updated as work progresses. It is not a claim of completed validation.
Sources & evidence
The evidence behind our argument
We keep our claims careful and our sources visible. These are some of the references that inform our mission and our standards.
Our case rests on a real and growing challenge, on recognised digital health evidence standards, and on the expanding science of digital biomarkers.
- Alzheimer's Society: dementia facts and figures
- NHS: how to get a dementia diagnosis
- NICE: Evidence Standards Framework for digital health technologies
- MHRA: software and AI as a medical device
External links open in a new tab. We also follow peer-reviewed research on speech, sleep, activity and other digital biomarkers in cognitive decline, and will reference specific studies as our own evidence base develops.
We are building an advisory group
We are forming a clinical and research advisory group to guide validation, ethics and responsible development. If you are a clinician, dementia researcher, data scientist or person with lived experience, we would value your perspective.
Advisory
Guided by the people closest to the problem
Responsible development is not something to do alone. We want clinical, scientific and lived-experience voices shaping DementiaDetect from early on.
Advisors help us keep claims proportionate, design studies well, protect against bias, and stay grounded in what families and clinicians actually need.
Explore a research collaboration
If you work in dementia research, health informatics or clinical evaluation, we would welcome a conversation.