Blood and Digital Biomarkers of Neurodegeneration for Dementia Diagnosis: UK Pilot Implementation and Health Economic Modelling

Supervisors: Dr Joseph Kwon; Professor José Leal

Aims: Timely and accurate diagnosis is a key dementia intervention strategy. It is crucial that the UK healthcare system leverages recent diagnostic innovations, including blood/digital biomarkers of neurodegeneration. This project aims to: analyse data from a UK-based pilot (SANDBOX) of triage using blood and digital cognition biomarkers of persons referred by general practitioner (GP) to memory clinic; and develop a health economic model evaluating the cost-effectiveness and health equity impact of national scale-up of the triage.

Methods: SANDBOX data will be analysed for diagnostic and prognostic values of blood (p-tau217, cell-free DNA), genetic (APOE4) and digital cognition biomarkers and for triage’s ability to reduce time from GP referral to diagnosis relative to standard UK practice (measured from matched controls identified in the Clinical Practice Research Datalink (CPRD)). A health economic model will be developed using diverse data sources, including the SANDBOX study data and other datasets such as CPRD and English Longitudinal Study of Ageing. Stakeholders including persons with lived experience of dementia and dementia diagnosis professionals/commissioners will inform the model development and analysis. Further UK datasets such as those from READ-OUT and DETERMIND studies will be used to validate the model. Cost-utility, distributional cost-effectiveness, probabilistic sensitivity and scenario analyses will evaluate the triage scale-up versus standard practice.

Expected outcomes: This project should yield key insights on implementation, diagnostic, and prognostic values of novel biomarker-based triage. Methodologically robust health economic model should inform policymaking to improve dementia diagnosis.

Preferred applicant background/skills: experience with health economic evaluation, preferably using decision-analytic models; strong quantitative research skills; knowledge of R.