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.

Become a research partner

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.

StageQuestionEvidence to generateStatus
Technical validationDoes the model perform reliably?Accuracy, sensitivity, specificity and bias testingIn progress
Retrospective testingCan signals identify known patterns?Consented dataset testing and model benchmarkingPlanned
Prospective studyDoes it work with real users over time?Longitudinal, real-world evaluationSeeking partners
Clinical pilotDoes it support workflow and decisions?Clinician feedback and pathway impactSeeking partners
Real-world evaluationDoes it improve outcomes or efficiency?Health-economic and safety dataFuture

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.

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.

Express interest in advising

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.