Dementia, mainly caused by Alzheimer’s disease (AD), is one of the biggest global public health challenges. It currently affects 50 million people, a figure which is expected to increase to 75 million by 2030.
Alzheimer’s pathophysiological changes start decades before symptom onset, thus opening a window for prevention. Gold standard AD biomarkers (PET and CSF) are invasive and expensive and thus not suitable for screening the general population.
- To develop a pre-screening tool based on a new family of machine learning algorithms to predict abnormality of core AD biomarkers from brain MRIs and reduce recruitment expenses by 50%.
Problem to Solve
Identifying cognitively unimpaired individuals who are at risk of developing Alzheimer’s and who could benefit from investigational therapies is challenging. Standard techniques to identify abnormal biomarkers are expensive and invasive. As a consequence, recruitment constitutes a major barrier to the investigation of preventive interventions for Alzheimer’s disease.
A family of machine learning algorithms has been developed to predict abnormal Alzheimer’s disease biomarkers from MRI scans at no additional cost. This is intended to be used as a pre-screening tool, to prevent individuals with a low likelihood of having abnormal biomarkers from undergoing expensive and invasive tests. Preliminary results show that, even though predictive power is limited, recruitment costs can be reduced to 47% and participant burden by up to 63%.