Alzheimer's: managing the conversation
Speech as an indicator of dementia
Researchers at the University of Southampton are breaking new ground in a bid to better understand the onset of Alzheimer’s, a disease that affects more than half a million people in the UK each year.
Investigating whether speech could be used as a reliable indicator for the progressive onset of Alzheimer’s and other forms of dementia, is the focus of the pioneering research by Anna Barney, Professor of Biomedical Acoustic Engineering and her team.
Anna explained: “Speaking puts a lot of strain on the brain. We have to work quite hard to speak and especially to take part in conversations, so we can tell how well the brain is working from looking at how people speak.”
In collaboration with Professor Peter Garrard based at St George’s Hospital Medical School, London, the team studied groups of patients shortly after diagnosis while they were talking to people who did not suffer from the condition.
Speaking puts a lot of strain on the brain. We have to work quite hard to speak and especially to take part in conversations, so we can tell how well the brain is working from looking at how people speak
By recording their conversations in a controlled environment, the team was able to identify some key differences in the way the two groups managed the conversation.
The team first identified what are known as ‘conversational overlaps’. These are points in conversations where people talk over each other, for example, by trying to interject or confirming what the other person is saying while they are speaking using words, noises or laughter.
By studying the conversations Anna found that people with an Alzheimer’s diagnosis overlapped a lot more than those without. “Alzheimer’s diagnosed speakers overlapped much more than healthy people did; they were managing the conversation less well and not judging as well when they could interrupt successfully. That can tell us that the planning part that happens in the brain during conversation is not working so well,” Anna added.
In addition the team also found patients managed the conversation differently when they talked to people they knew when compared to talking to strangers. This ‘familiarity factor’ could prove crucial in increasing understanding of how a patient behaves in clinic compared with how they behave at home.
Anna explained: “We found that Alzheimer’s patients did more confirming of the other speaker. It appeared they were finding it more difficult to manage turn taking and so used the confirming to stay in the conversation without tackling the more difficult task of taking a full turn.
“That means that what is seen in the clinical setting might not be the same behaviour that is being displayed at home. This is important as it means that if we are to use speech as an indicator for the progression of the condition, then we need to find a way of measuring this at home.”
Alzheimer’s diagnosed speakers overlapped much more than healthy people did; they were managing the conversation less well and not judging as well when they could interrupt successfully. That can tell us that the planning part that happens in the brain during conversation is not working so well
Anna has also begun extending the research to investigate whether monitoring how prevalent repeated speech is can be an identifiable symptom of the onset of the condition. Anecdotal evidence from carers and relatives suggests those with the condition often repeat a particular phrase, which again might not be evident when the patient presents at clinic.
Anna explained how the challenge was then to gather these samples of speech while the patient was in their own home, but also taking into account other factors. The team was faced with considering issues surrounding invading the privacy of the patient or other people they might be interacting with, how efficient the task would be in recording lengthy samples of speech, and any distress that might be caused to the patient who was having to wear the recording device.
In order to collect the samples in the least intrusive way researchers used smaller and more sensitive accelerometers, which measure vibrations and are used in Eurofighter technology. Having recorded the samples researchers were then able to design a computer programme capable of picking out repeated phrases without someone having to listen to hours of recordings, making the task a lot more efficient.
The work has already led to promising results and it is hoped that it could lead to researchers being able to analyse the speech of Alzheimer’s patients in their own homes where a more accurate picture of how the condition is progressing, using their speech, can be assessed.
Anna was one of a number of speakers at a Living with Dementia conference hosted at the University in partnership with Alzheimer’s Research UK.