British researchers have developed an AI model that could predict Alzheimer's disease much earlier

Researchers at the University of Cambridge and other institutions have developed a new artificial intelligence (AI) model that can not only eliminate expensive or complex tests, but also predict Alzheimer's disease earlier.

An AI prediction model led by the University of Cambridge provides a non-invasive and low-cost way to effectively predict whether a study subject will develop Alzheimer's disease in the next three years, Xinhua reported. The study has been published in the British journal Electronic Clinical Medicine.

Based on data collected from cognitive tests and MRI scans of 400 patients with brain atrophy, the research team used machine learning algorithms to build an AI prediction model and tested it using real data from multiple clinics in the UK, Singapore and elsewhere. Due to the use of multimodal data such as text and pictures, the model can more accurately predict the probability of early disease transforming into Alzheimer's disease than traditional clinical diagnosis.

The results showed that the model was 82 percent accurate in identifying people who would develop Alzheimer's within three years and 81 percent accurate in identifying people who would not develop Alzheimer's within three years.

At present, accurate diagnosis of Alzheimer's disease generally relies on expensive or complex tests, such as lumbar puncture or tomography, but these tests are not available in all healthcare facilities, so up to a third of patients may be misdiagnosed, and some patients are diagnosed too late to receive effective treatment.

More than 55 million people worldwide suffer from dementia, the most common type being Alzheimer's disease. In the future, the research team hopes to expand the model to predict other types of dementia and use different types of data for detection, such as markers in blood tests.