blog

AI is a rapidly evolving field, with huge potential. Though most of us think of ChatGPT and AI artwork when we consider this buzzword of the 2020s, AI is showing huge potential in healthcare, with a promising prognosis for dementia care and treatment.

AI in dementia diagnosis

Prompt and accurate diagnosis of dementia is key to achieving the best treatment response, but achieving this is notoriously difficult. The ‘Cognospeak’ AI system, from the University of Sheffield, aims to change that.

Cognospeak analyses a patient’s speech and language patterns and uses this data to detect memory deficits, including Dementia and Alzheimer’s Disease. These patterns can show early indicators of cognitive decline which may be missed in traditional assessments. Furthermore, Cognopeak has shown 90% accuracy in distinguishing cognitively healthy people from those with Alzheimer’s.

Having shown promising initial findings, Cognospeak is now undergoing further trials in memory clinics nationwide. It is hoped that the tool will result in faster diagnosis of dementia; therefore increased treatment efficacy. Furthermore, it may reduce the need for hospital-based pen and paper tests, which cause patients undue anxiety.

Further information about Cognospeak can be found at:

• AI tool could speed up dementia diagnosis (University of Sheffield)

AI and independence

The loss of independence resulting from dementia is frustrating for those living with dementia and upsetting for their loved ones. However, AI may be able to prolong the time which a person with dementia can live independently and quell some of the worries of their loved ones. The UK Dementia Research Institute have developed the ‘Minder’ project, in collaboration with Imperial College London. Minder is a smart home care system which aims to promote independent living. A person living with dementia is given wearable sensors and environmental sensors are placed around their home, which connect to a Minder app. The app sends sensor data to a monitoring team, consisting of AI to monitor trends in a person’s behaviour and to a clinical team. The AI remotely detects signs a person is at risk in their home, for example changes in walking that could increase the risk of a fall (via motion sensors), or increased wandering (via door sensors). Smart plugs may also be used to monitor use of appliances related to eating and drinking, such as kettles and microwaves. The clinical monitoring team then inform the individual’s carer according to their contact preferences.

The multidisciplinary ‘Minder’ team are currently developing augmented hearing aids which can continuously monitor a patient’s brain activity and heart rate, which could contribute to the wearable sensor technology. Furthermore, balance could be monitored via these sensors to indicate fall risk and audio cues could be provided through the device to support independence. Currently, the hearing aid is being revised in healthy volunteers prior to use in dementia patients, and the Minder app is undergoing early trials with dementia patients and their families.

Further information on ‘Minder’ can be found at:

• Minder meeting place
• UK DRI Care Research & Technology (UK Dementia Research Institute)

AI and prognosis

The progression of dementia may also be predicted with AI. Several studies have explored the use of AI to predict mild cognitive impairment (MCI) progression to Alzheimer’s Disease. These studies used data collected using imaging techniques, such as PET and MRI and EEG, which records brain activity. Large datasets from the Alzheimer’s Disease Neuroimaging Initiative (ADNI), who collect large amounts of data from Alzheimer’s, MCI patients and healthy elderly adults over many years, make the development of AI databases for such applications possible. For example, a team at the Alan Turing Institute developed an AI trained to detect changes in brain structure which, combined with a patient’s memory test results, predicted the chance of a patient’s MCI progressing to Alzheimer’s. The algorithm could also predict the rate of cognitive decline in an individual patient. In some cases, the algorithm correctly predicted an individual with no symptoms of cognitive decline would develop Alzheimer’s. An accuracy of 80% was obtained for prediction of progression to Alzheimer’s. It is hope that the AI can be trained to distinguish the different types of dementia, based on the different patterns in the loss of brain tissue they present. Clinical trials using this AI approach are now taking place in centres including Cambridge and Brighton.

More information can be found at:

• AI could detect dementia years before symptoms appear (University of Cambridge)