Brain Sciences

New approaches to early diagnosis, monitoring and treatment of cognitive and mental health issues in neurological and psychiatric disorders are urgently needed. In particular new approaches are necessary to deliver at the scale demanded. Delivering this care is especially difficult in our London catchment area because of high rates of co-morbidity, social deprivation and social inequalities. We will develop new approaches in two underserved populations in this area: the old and the young with a dual focus on late-life cognitive impairment and on emotional and behavioural dysregulation seen across psychiatric disorders in young people. This approach is likely to deliver a substantial impact as around 25% of NHS beds are currently occupied by patients living with dementia and by 2040 those affected by dementia will increase to around 1.6 million in the UK. In addition, mental illness is experienced by one in four people in the UK during their lifetime, with particular risks seen during adolescence.

• Develop new ways to proactively manage health and social care needs for people living with dementia via novel engineering and technology approaches to dementia care.
• Develop new methods to identify and monitor those at risk of and with dementia via digital tools for remote precision assessment of behaviour, cognition and mental health.
• Develop novel bioelectronic medicine approaches to treating cognitive impairment and dementia (such as non-invasive brain stimulation).
• Improve clinical outcomes in children and young people with emotional and behavioural dysregulation.
• Investigate the potential of treating psychiatric disorders with psychedelics and characterize addiction neurobiology (substance, gambling) and obesity for a personalised medicine treatment approach.
• Develop and deploy the computational infrastructure for analysing large-scale clinical data and pilot digitally enhanced clinical decision-making.
• Extend a real-world test bed designed to support the efficient evaluation of new approaches to health and social care.

Detailed objectives can be found here

1. We are evaluating novel engineering and technology approaches to dementia care, using the home monitoring platform Minder. Professor David Sharp’s team are testing smart sensors that can collect data about a person’s behavioural patterns in the home by using artificial intelligence to make predictions about what that person might do. They will use this information to develop ways to support them in their home as their dementia progresses.

2. Dr Martina Di Simplicio’s team is investigating the effectiveness of a new intervention called the IMAGINATOR to diagnose and treat self-harm in young people. IMAGINATOR trains people to develop and use more positive behaviours as an alternative to self-harm when dealing with distressing emotions. They are also testing a novel and specially designed app to help reduce symptoms which can be used after the therapy has finished.

Comparing Patient Symptoms, Brain Nerve Pathways and Markers of Dementia in Normal Pressure Hydrocephalus

What is this project about?

Normal Pressure Hydrocephalus (NPH) is a common but frequently missed condition in older adults, where a build-up of fluid in the brain causes problems with walking, thinking, and bladder control. Unlike most dementias, NPH can be treated with a surgical procedure called “shunting”, which drains the excess fluid. However, many thousands of people in the UK are likely living with undiagnosed NPH, mistakenly told they have Alzheimer’s or another dementia. This project aims to identify better ways to distinguish NPH from other dementias and to determine which patients are most likely to benefit from surgery.

Why does it matter?

Getting the right diagnosis for NPH can be life-changing — surgery can restore a person’s ability to walk, think clearly, and regain bladder control. Without accurate diagnosis, patients miss out on a treatable condition and face unnecessary decline in quality of life. Current clinical tools are not reliable enough to tell NPH apart from other dementias, and there is an urgent need for better biological, imaging, and cognitive markers. This project is directly addressing that gap by developing and testing a range of new diagnostic approaches across multiple patient groups.

What are the outputs of the project?

The project has developed a new digital gait assessment tool that measures how patients walk, proven superior to standard NHS clinical measures and now being published in Practical Neurology. Working with Prof David Sharp at Imperial College, MRI brain imaging has been completed in 13 participants, with repeat scans planned after shunting to track how brain nerve pathways change with treatment. A new digital cognitive test battery — the NPH-Cognitron — has been developed and tested in collaboration with Professor Adam Hampshire at King’s College London, with promising results that could establish a standardised assessment tool for NPH research worldwide. Blood dementia biomarkers are being analysed with Heinrich Zetterberg and Amanda Heslegrave at the MRC Dementia Biomarker Factory, CSF synuclein testing is underway with Dr Marcello Barria at Edinburgh University, and whole genome sequencing of blood samples is being conducted through the Global Parkinson’s Disease Genetic Programme. In a world first, brain tissue from a living NPH patient was analysed using novel electron microscopy techniques at the Roslind Franklin Institute with Prof Michael Grange, directly visualising nerve cell microstructures in the ageing brain. Two publications are in progress, and a UK-wide NPH Registry has been established through the Association of British Neurologists to better understand the scale of patient need and treatment timelines.

How were patients and the public involved?

Patients and carers have been involved from the outset, with the study protocol reviewed by an ICHT NPH patient and carer cohort. A survey of 40 NPH patients and carers through the SHINE network revealed that faster and more accurate diagnosis and treatment were their top priorities — findings that directly shaped the research focus. Patients also advised on consent procedures and contributed to the development of quality-of-life measures, highlighting that existing generic tools do not adequately capture the NPH experience. This has led to the development of a dedicated quality of life assessment tool for NPH, currently forming the basis of a NIHR doctoral fellowship application.

Improving How We Assess Thinking and Communication Skills After Stroke

What is this project about?

Stroke can affect a person’s ability to think, remember, and communicate, yet the tools currently used to assess these difficulties are often poorly suited to stroke survivors. This project developed a new online cognitive testing tool — called IC3 — specifically designed to assess memory, attention, and language in people affected by stroke and vascular dementia. The tests use cutting-edge artificial intelligence to ensure they are fair and accessible, including for people with speech difficulties. New speech recognition tools were also developed to automatically detect language problems and track recovery over time.

Why does it matter?

Stroke is one of the leading causes of long-term disability in the UK, and difficulties with thinking and communication are among the most common and debilitating consequences. Existing assessment tools are often too complex, require a clinician to administer, and are not designed with stroke patients in mind. By making cognitive and language assessment available online and unsupervised, this project enables patients to be tested more frequently, more conveniently, and from home — providing a much richer picture of recovery over time.

What were the outputs of the project?

The IC3 tool has been developed and validated in collaboration with Professor Adam Hampshire at King’s College London, building on an established platform to create stroke-specific assessments. A thematic analysis of structured feedback from over 100 patients who have used the tool is underway, informing further refinements. The project has secured significant further funding, including an MRC Transition Award of £405,509 to support a longitudinal study of cognition after stroke using digital health technology, and a £71,396 UKRI Impact Acceleration Award for speech recognition validation in aphasia. The team has also been appointed Co-Director and Imperial lead for a new Doctoral Training Centre for Vascular and Immune contributors to dementia, funded by Alzheimer’s Society at approximately £3 million. A clear and accessible Results page has been co-designed with patients and healthcare professionals to ensure findings are meaningful and easy to understand.

How were patients and the public involved?

Patients have been central to the development of this project throughout. A series of one-to-one feedback sessions with stroke patients directly shaped the design of the Results page, ensuring it communicates findings in a clear and accessible way. An educational workshop was delivered to the Mosaic Trust to improve understanding of stroke and increase awareness of research participation in underserved communities in London. The team has also conducted a thematic analysis of feedback from over 100 patients who have completed the IC3 tests, with findings being prepared for publication.

Privacy-Conscious AI Model Development for Personalising Multiple Sclerosis Care

What is this project about?

Multiple sclerosis (MS) is a long-term condition affecting the brain and spinal cord, and finding the right treatment at the right time for each individual patient remains a major challenge. This project developed and deployed artificial intelligence (AI) models across two NHS hospitals — Hammersmith Hospital and the National Hospital for Neurology and Neurosurgery — to improve how MS is monitored and treated. Crucially, the AI was built using “edge computing” and “federated learning”, meaning patient data never had to leave the hospital, protecting privacy while still allowing the AI to learn from thousands of real patient scans.

Why does it matter?

MS disproportionately affects people from minority ethnic backgrounds and those with other health conditions, yet these groups are routinely excluded from clinical trials, meaning AI models trained on traditional datasets may not work fairly for everyone. By deliberately including patients from diverse backgrounds and incorporating social deprivation data, this project took meaningful steps towards equitable precision medicine — ensuring AI tools work for all patients, not just those typically represented in research. The goal is to help doctors prescribe the right drug to the right patient at the right time, delaying disability and improving quality of life.

What are the outputs of the project?

A functioning federated AI infrastructure has been successfully installed and operationalised at Hammersmith Hospital, processing brain and spinal cord MRI data from over 3,200 MS patients across more than 6,000 scans. Three AI models are being validated across both hospital sites: one automatically measuring MS lesions and brain structures from MRI scans, one identifying biologically distinct MS subtypes, and one predicting disease activity using multimodal data including MRI and demographic information. Performance across both hospitals was comparable, confirming the feasibility of training AI models without sharing sensitive patient data between sites. A new working collaboration has been established between University College London Hospitals and Imperial College Healthcare NHS Trust.

How were patients and the public involved?

A Patient and Public Involvement (PPI) steering committee of four people with lived experience of MS and caring was convened, representing a range of ages, genders, and ethnic backgrounds. Their input directly shaped the study’s inclusion criteria, leading the team to remove upper age limits, explicitly include people with common comorbidities, and monitor representation across ethnic groups. The committee advised on how to present results in a patient-friendly format — focusing on what predictions mean for quality of life and treatment choices, not just technical metrics — and recommended inclusive imagery and communication materials. They also endorsed the team’s plan to scale the federated infrastructure UK-wide, recognising that broader participation would improve fairness and generalisability for all people living with MS.

Predicting Risk of Falls in Early-Stage Dementia or Parkinson’s Disease Using Remote Monitoring of Daily Life Activity

What is this project about?

Falls are a serious and common problem for people living with dementia or Parkinson’s disease, often leading to hospitalisation and a rapid decline in independence. This project investigated whether data from everyday wearable devices — such as Fitbits — could be used to predict who is at risk of a serious fall, without the need for clinical assessments. Using data from over 400 people in the UK Biobank, the team explored the relationship between physical activity patterns, brain scan findings, and fall risk in people with early-stage dementia. An AI model was also developed to identify patterns in activity data that may signal increased fall risk.

Why does it matter?

Falls in people with dementia are a leading cause of hospital admissions and can mark a turning point in a person’s condition. Being able to predict fall risk early — using data that people already generate through everyday wearables — could allow earlier intervention, reduce harm, and support people to live independently for longer. This project found that around 31% of people received their dementia diagnosis around the time of their first serious fall, suggesting that falls may be an early warning sign of neurodegeneration that is currently being missed.

What are the outputs of the project?

Screening of 403 UK Biobank participants revealed that nearly half experienced a fall within six months of their dementia diagnosis, highlighting the close relationship between neurodegeneration and fall risk. Analysis showed that physical activity data alone was not a reliable predictor of falls, but that brain scan information — particularly white matter changes — added meaningful predictive value. A prototype AI model was developed using a novel self-supervised learning approach, achieving approximately 80% accuracy on a small dataset of 200 people, demonstrating its potential for use with larger datasets in future. The project findings have been aligned with the NIHR BRC Brain Sciences theme and will contribute to the broader Minder home monitoring platform for dementia patients.

How were patients and the public involved?

A PPIE session was organised specifically for people living with neurodegeneration, focusing on individuals with dementia, to understand their views on using wearables and AI for health monitoring. Participants expressed a clear preference for more personal, human-centred approaches to health monitoring, rather than fully automated digital systems — feedback that directly informed how the team framed its findings and future research direction. The team also presented its work to NIHR BRC community partners, who highlighted the importance of identifying which specific types of physical activity are most protective against falls, a consideration that has been incorporated into future research planning. These insights reinforced the need for tools that complement, rather than replace, meaningful human involvement in care.

New AI Tools for Brain Image Analysis

What is this project about?

Brain scans are one of the most important tools doctors have for diagnosing and monitoring conditions such as brain tumours and neurological diseases. However, analysing these scans accurately and efficiently remains a significant challenge, particularly when scans are taken using different imaging techniques. This project developed advanced AI models capable of analysing brain scans across multiple imaging types — including T1, T2, and FLAIR contrasts — to automatically identify and measure brain lesions and tumours. The AI can produce detailed maps showing the location and volume of abnormal tissue, providing clinicians with more consistent and objective information to support diagnosis and treatment planning.

Why does it matter?

Accurate measurement of brain lesions and tumours is critical for understanding how a disease is progressing and whether treatment is working. Currently, this analysis is time-consuming, subject to human variability, and often requires multiple high-quality scans that are not always available. By developing AI that works reliably across different scan types and clinical settings, this project makes high-quality brain image analysis more accessible — including for NHS patients whose scans may not meet the standards typically required by research tools. This could ultimately speed up diagnosis, improve monitoring, and support better treatment decisions for patients with brain tumours and neurological conditions.

What are the outputs of the project?

The project has produced four peer-reviewed publications in leading international venues. The AI models developed include a foundation model for brain lesion segmentation that handles multiple imaging modalities, a model for brain tumour segmentation using a novel two-stage image synthesis approach, and a tool for segmenting heterogeneous brain lesions with anatomical constraints. Digital infrastructure has been established within the NHS Trust to receive brain scan data and analyse brain volumes using AI, enabling the feasibility of deploying these tools in real clinical imaging datasets to be assessed. An academic collaboration with Prof Yaou Liu, Neuroradiologist at Capital Medical University in Beijing, has supported brain image analysis and co-authorship of publications.

How were patients and the public involved?

The project was presented at a BRC Community Partner Meeting, where the team discussed the need for better tools to interpret brain images and gathered feedback from community partners. This engagement helped contextualise the research within the real-world needs of patients and the public, reinforcing the clinical relevance of developing AI that can work with routine NHS imaging data rather than only research-grade scans. The team remains committed to ensuring that the tools developed are ultimately accessible and beneficial to NHS patients.

We aim to reduce the current inequalities of access of our most deprived communities to involvement in research. We will do this by appointing community partners who have strong relationships with local communities in North West London. We are working with the Imperial Patient Experience Research Centre (PERC) to promote involvement and participation in research from underrepresented communities, particularly in collaboration with the two associated mental health trusts, West London NHS Trust and Central and North West London NHS Foundation Trust.  We will provide PPIEP training for the researchers in the Theme and the community partners who have an interest in working with us. We are working on building our Strategic PPIEP Steering Group which will attend our Theme management meetings to report on progress against our PPIE Strategy and ensure PPIEP is addressed in governance.

In addition, in collaboration with the Helix Centre, the UK Dementia Research Institute Care Research and Technology Centre team we have nominated Minder Champions, a group of both carers and people living with dementia who regularly participate in PPIEP activities to help co-design and understand the acceptability of the new technologies developed by the centre to support people living with dementia and their carers.

In our Theme, we seek to improve equality, diversity and inclusion (EDI) in all aspects of our work. We will work closely with the BRC EDI oversight committee to adhere to the BRC EDI strategy and apply the EDI principles to our recruitment, hiring, and retention practices as well as our community engagement practices. One way we intend to demonstrate improvements towards EDI principles will be by recording and monitoring recruitment metrics of the new members joining our teams and of those who apply and are awarded funding to deliver our research objectives. Also, during the selection process of our community partners, we have taken into account different ethnic groups and other demographics representative of the wider community to ensure that the views of the diverse local community in the West London area are represented in our research work. We also seek to support young emerging researchers in our Theme and create equitable opportunities for future leaders in various research fields.

Jacob D. King et al. The association of severe COVID anxiety with poor social functioning, quality of life, and protective behaviours among adults in United Kingdom: a cross-sectional study. BMC Psychiatry volume 23: 117 (2023). DOI https://doi.org/10.1186/s12888-023-04595-1

Meg J. Spriggs et al. Study Protocol for “Psilocybin as a Treatment for Anorexia Nervosa: A Pilot Study”. Front. Psychiatry, Vol 12 – 2021. DOI : https://doi.org/10.3389/fpsyt.2021.735523

Robin Carhart-Harris et al. Trial of Psilocybin versus Escitalopram for Depression. N Engl J Med. 2021 Apr 15;384(15):1402-1411. DOI: 10.1056/NEJMoa2032994.

In accordance with the Imperial BRC’s PPIEP Strategy , our theme has recruited a group of Community Partners to act as critical friends to our theme and share their valuable lived experience with our researchers and health professionals to help improve the relevance and quality of our research for the benefit of our North West London population.

“I am a mental health, media and marketing professional. I work for a range of organisations within mental health, including Mind, the Anna Freud Centre and The Royal College Of Psychiatrists. I bring a cultural lens to my work and am a passionate advocate for increased awareness and support around neurodiversity.” Abiola Awojobi-Johnson

“I am a student at the University of Bristol involved in the Brain Science theme at Imperial, I am very excited to be a Community Partner and hope to improve local initiatives and do what I can to improve health services.” Ellie Roberts

Siobhán McGrath: I spent my life working both on a voluntary basis and professionally with vulnerable adult and older people with learning/ physical disabilities, addiction issues and mental health issues including dementia.