AI model and cognitive test combine to predict Alzheimer's
A cognitive test currently used to diagnose dementia and other cognitive illnesses has been adapted to enable Alzheimer's prediction.
A study carried out by IBM Research and Pfizer developed an artificial intelligence model which uses natural language processing to analyse people's written responses to the Cookie Theft test, where they describe the scene in a picture.The model then detects subtle changes to language such as grammatical errors and sentence structure in people's responses over time.
The paper, published in the Lancet eClinical Medicine, reported that the AI model could be used to predict the likelihood of an individual going on to develop Alzheimer's up to seven years in advance and with a 70% accuracy.
To find out more, and to read Alzheimer's Society's response to the findings, read the BBC Health article here, or find the clinical research paper here.
Related articles
Promoting prevention, supporting management
Led by proactive clinicians determined to see improvement in the way we prevent, diagnose and manage dementias, Dementia Academy supports healthcare professionals with the latest tools, resources and courses to do just that.