Labhpreet Kaur is a PhD student at the University of Sheffield looking to improve the diagnosis of dementia and memory problems. She is carrying out research on how we can use AI to include experiences of family members and caregivers to improve diagnosis of dementia.

She is part of the research group that has developed a digital tool called Cognospeak™. The tool uses artificial intelligence (AI) to analyse people’s speech and language. It identifies those who are at risk of developing memory problems or dementia. Labhpreet shares her insight into how AI is being developed for dementia research. 

What is the study investigating?

“This study aims to find out how information provided by family members and caregivers can improve how dementia is diagnosed. Family members and caregivers of people with Mild Cognitive Impairment (MCI) and dementia take part in a short online interview to share the early changes they noticed in their relatives before diagnosis. 

“We will then use large language models, which are a type of artificial intelligence. This will automatically analyse these caregiver accounts. It will help us explore how well AI can summarise this information. It will also identify signs of memory problems and dementia. With there being the possibility of diagnosis based on caregivers’ observations. 

“We will compare the AI’s responses with assessments from consultant neurologists. This will help us understand how accurate the AI is. We will also look at whether it could support doctors in making quicker diagnoses.”

How are AI tools being developed for dementia research?

“AI is being developed across many areas of healthcare. From analysing brain scans and improving patient care to supporting clinical training. In dementia research, AI is being used to detect early signs of the condition. It analyses complex real-world data such as speech, language and behaviour.

“At the University of Sheffield, our team has developed CognoSpeak™. The software is led by Dr Daniel Blackburn, Clinical lead of CognoSpeak™ and Senior Lecturer and Honorary Consultant Neurologist at the University of Sheffield, and Professor Heidi Christensen, Technical lead of CognoSpeak™ and Professor in Computer Science at the University of Sheffield.

“It is a digital tool that uses a virtual agent to ask people with suspected memory problems questions about their memory and health. It uses AI to analyse their speech. Early research has shown that CognoSpeak™ can help identify subtle changes that may show early cognitive decline. It is offering a quick and accessible way to support diagnosis.

“We now want to take this further by developing a version of the tool focused on carers. This new feature will ask family members and caregivers about the changes they’ve noticed in the person they care for. Then it will use AI to summarise this information. This is helping to build a fuller picture of their relative’s health. In the future, this could make tests more accurate. It could also reduce waiting times and ease pressure on memory services by enabling more efficient, remote screening.”

Why is it important to start using AI tools in research for dementia?

“Dementia is one of the biggest health challenges we face, and diagnosing it early can be difficult. Many symptoms, such as changes in speech, memory or behaviour, can be subtle at first and vary from person to person. Traditionally, memory tests rely on pen-and-paper screening tools. They are useful but can sometimes miss the early or more subtle signs of cognitive decline. At the same time, memory clinics and diagnostic services are often overstretched. This is leading to long waiting times and delays in getting support.

“AI offers a powerful way to help with these challenges. It can analyse large amounts of complex information. This will detect early changes that might not be picked up through standard tests. By learning from patterns in this data, AI tools can support clinicians in making more accurate and consistent diagnoses. Introducing AI into dementia research can also make testing more accessible. For example, tools like CognoSpeak™ can be used remotely, allowing people to take part from home. This can save time for both patients and clinicians. It can reduce waiting lists and help people showing early signs of dementia are supported sooner.”

What are you hoping to achieve from this study?

“By comparing the AI’s summaries with assessments made by clinicians, we aim to see how well AI could help identify signs of dementia. In the long term, this research could lead to digital tools that support doctors in making quicker and more accurate diagnosis. This will help to reduce waiting times, and ease pressure on memory services. It will also offer remote access to assessments for people who may find it difficult to attend clinics in person.”

How has Join Dementia Research helped to support your study?

“Join Dementia Research has been an invaluable resource for our study. It is helping us connect with people who have experienced dementia. They are also passionate about contributing to research that can make a difference. It has helped us reach a diverse group of participants across the UK. This ensures that the voices of real caregivers are central to our work. Using Join Dementia Research has also made recruitment quicker, more inclusive and effective. This is allowing us to match the right participants to our study.”

What would you say to other researchers about using Join Dementia Research?

“I’d strongly encourage other researchers to use Join Dementia Research. The platform makes recruitment faster, easier and more inclusive. It is helping you connect with motivated and engaged participants.

“The Join Dementia Research team themselves have been incredibly responsive. They are supportive and make the whole process straightforward from start to finish. For any researcher working in this field, Join Dementia Research is an invaluable platform for connecting with the people who matter most – those with lived experience.”

Could you use Join Dementia Research to help recruit to your study? Visit our ‘Researchers’ page for more information.