The brain cell degeneration and death that lead to Alzheimer disease begin 10 to 20 years before the very first symptoms appear. But what if experts could predict a person’s risk of developing the disease up to five years earlier? Mallar Chakravarty, researcher at the Douglas Hospital Research Centre affiliated with McGill University, is taking up that very challenge.
Mallar Chakravarty believes that, with early detection, patients could slow the disease’s progression by changing their lifestyles.
Along with colleagues at the University of Toronto, he has set his sights on artificial intelligence to anticipate whether an individual’s cognitive abilities are unlikely, likely or very likely to progress to Alzheimer’s disease over the next five years.
The researchers developed an algorithm that integrates clinical data from 800 people—the outcomes of cognitive tests, information on the presence of genotypes that make an individual four to eight times more likely to develop the disease and imaging results. Based on the dataset, the algorithm is trained to detect the signs of cognitive decline by pinpointing the disease’s subtle signature in the brain and behind the clinical findings.
The algorithm was successfully tested as part of an Australian cohort study. The new clinical data on the subjects was entered into a software program, and the algorithm was then able to model each person’s own path to Alzheimer disease. The scientists plan to repeat the process with a group of 500 Quebecers experiencing memory decline. Follow ups are planned 4, 6 and 10 years later to validate the accuracy of the predictions.
Mallar Chakravarty believes that, with early detection, patients could slow the disease’s progression by changing their lifestyles: getting more physical and cognitive exercise, eating better or not drinking alcohol.
Convinced of the potential of their method, the researchers are in discussions with a partner interested in developing a user-friendly tool for physicians in their clinical practices.