Decoding death and dying in people with dementia by digital thanotyping: a longitudinal observational study protocol

Introduction As individuals approach the end of life, physical, mental and social functions decline. However, reliable methods to detect and characterise symptoms of the terminal phase in people with dementia are lacking. Advances in non-invasive sensor technologies and oral sampling methods offer new opportunities to improve assessment of this phase. Methods and analysis This is a 5-year longitudinal observational study using a mixed-methods and multidisciplinary approach. The project, Decoding Death and Dying in people with Dementia by Digital thanotyping (5-D)—with ‘thanotyping’ derived from the Greek thánatos (death) and phaino (to reveal)—aims to develop methods and tools to accurately characterise the dying process in people with dementia. The study integrates clinical assessments with advanced sensing technologies to monitor pain and distressing symptoms. Oral examinations will be conducted with biomaterials from the oral cavity. Prediction model development will draw on dynamic systems theory, signal processing, artificial intelligence with a human-in-the-loop approach and statistical modelling to identify patterns and trajectories associated with end-of-life processes. A sample of approximately 270–300 participants is required to obtain at least 95 end-of-life trajectories. While focusing on established symptoms of the dying phase, some phenomena may be difficult to capture due to the complex and multifactorial nature of the dying process. The use of digital modelling to predict end-of-life trajectories raises important ethical considerations. These include challenges in obtaining valid informed consent from individuals with impaired decision-making capacity, potential risks of psychological harm and compromised dignity, and the possibility that predictive output may inadvertently influence clinical decision-making in a self-fulfilling manner. Further concerns include uncertainty and bias in model predictions, as well as issues of data privacy, ownership and governance. The study emphasises transparency in model development, minimisation of participant burden and appropriate clinical oversight. Ethics and dissemination The 5-D study has been approved by the National Committee for Medical and Health Research Ethics (NEM 2023/166) and the Regional Committee for Medical and Health Research Ethics (REC Vest; 657596), including the approval for storage of saliva and dental plaque samples in a biobank at the University of Bergen, Norway. Findings will be disseminated through peer-reviewed publications, conference presentations and engagement with stakeholders, including clinicians, families and policymakers. Trial registration number NCT06437132.