How it works
Multiple signals, brought together with care
Dementia does not usually appear in one single moment. It often emerges through small changes over time. DementiaDetect is being designed to listen for those changes across multiple signals, and to do so responsibly.
Multimodal by design
The signals we are exploring
No single signal tells the whole story. The aim is to combine many everyday signals, so subtle and gradual change is easier to recognise than it would be from any one source alone.
Speech and language
Patterns in fluency, word-finding and how language is used over time.
Sleep patterns
Changes in rest and sleep that can be relevant to brain health.
Physical activity
Movement, steps and changes in daily routine and function.
Cognitive tasks
Structured, friendly check-ins designed to track change over time.
Daily routines
Shifts in everyday behaviour that families often notice first.
Wearable and device data
With consent, signals from everyday devices people already use.
Self-reported information
What people and families share about how things feel.
Clinical information
Where appropriate, context from existing care, with consent.
Every signal is used only with consent, and you stay in control of what is shared.
Speech · Sleep · Steps · Signals
From everyday data to earlier insight
Capture, with consent
Relevant signals are brought together only with clear, informed consent and control over what is shared.
Detect patterns over time
AI looks at how signals change, helping surface patterns that may be associated with cognitive change.
Support earlier action
Insight is framed to support conversations with healthcare professionals, not to deliver a diagnosis.
Human oversight
AI surfaces patterns. People make decisions.
AI can identify patterns across complex data that may be difficult to notice manually, especially when signals are subtle, gradual or spread across different behaviours.
That capability is powerful, which is exactly why we treat it carefully. DementiaDetect is designed to be assistive. It helps surface patterns that may support earlier conversations, assessment and care. It does not make clinical decisions, and it is not a diagnostic device.
If you are worried about memory, confusion or cognitive change, please speak to a qualified healthcare professional. DementiaDetect is being developed as a support tool and is not a substitute for medical advice, assessment or diagnosis.
Where insight is shown, our aim is to make it understandable: what kind of signals informed it, and what a sensible next step might be.
Clear boundaries
What DementiaDetect does, and does not do
It is designed to
- Bring together multiple signals to look for meaningful patterns
- Support earlier awareness and more informed conversations
- Help indicate when further assessment or support may be helpful
- Provide longitudinal monitoring that respects consent and privacy
- Keep clinicians and people firmly in control of decisions
It does not
- Diagnose dementia or any medical condition
- Replace clinical judgement or professional assessment
- Promise prevention, cure or certainty
- Present a risk insight as a medical conclusion
- Make decisions on a person's behalf
In practice
What using DementiaDetect could look like
The product is being developed, so the journeys below are illustrative. They show the experience we are designing for families and for clinicians.
For a family member
Complete gentle brain-health check-ins
Short, friendly check-ins designed to be calm and easy, not a test to fear.
Connect approved devices or share signals
With consent, link everyday devices or share relevant information. You stay in control.
Track changes over time
See how signals such as speech, sleep and activity change across weeks and months.
Receive clear, non-diagnostic insight
Plain-language insight about patterns that may be worth attention. Never a diagnosis.
Know when to seek support
Clear guidance on when it may help to speak to a healthcare professional.
Share a summary, where appropriate
Choose to share a concise summary with a clinician to support the conversation.
Change over time
Last 12 weeksExample insight. A small change in language patterns over recent weeks. It may help to discuss this with a healthcare professional.
Illustrative preview only. Not a diagnosis and not a real result. Final design and outputs are being developed with clinical input.
What a clinician could see
- A consented, longitudinal summary of signals over time
- Change highlighted across multiple signals at once
- Plain-language, non-diagnostic risk insight
- Support for prioritising who may need further assessment
- A concise summary that can be exported into the pathway
Designed to support, not replace, clinical judgement.
For a clinician
Receive consented signal summaries
Longitudinal, consented summaries that respect privacy and control.
See change over time
A clear view of how multiple signals are moving, not a single snapshot.
Review non-diagnostic insight
Risk insight framed to inform judgement, never to deliver a diagnosis.
Prioritise assessment
Support for identifying who may benefit from further assessment sooner.
Export a concise summary
Bring a brief, structured summary into existing care pathways.
See where this is heading
Explore our research focus and validation roadmap, or join the early access community to help shape responsible development.