Alzheimer’s disease is a terminal neurodegenerative disease that has historically been diagnosed based on observing significant memory loss. There is currently no cure or disease-modifying therapy, despite hundreds of clinical trials.
It is thought these trials may have a high failure rate because the people enrolled are in the last stages of the disease, likely already suffering a level of brain tissue loss that cannot easily be repaired.
Thus, researchers have put their mind to how to detect this disease earlier, while a chance may still exist to slow its progression.
Recent research has shown a biological marker associated with the disease, a peptide called amyloid-beta, changes decades before any memory-related issues are apparent. Examining the concentration of the peptide in a person’s spinal fluid can provide an indication of Alzheimer’s disease risk decades before any memory-related issues are apparent.
IBM researchers released a paper this week detailing their work in employing machine learning to identify a set of proteins in blood that can predict the concentration of amyloid-beta in spinal fluid.
That is, the work could one day help clinicians predict the risk of Alzheimer’s in their patients, decades prior to any memory effects, through a simple and routine blood test.
The study isn’t the only one exploring a blood test for Alzheimer’s, but it is the first to use a machine learning approach to identify sets of proteins in blood that are predictive of a biomarker in spinal fluid.
The research is still in its early phases, but could potentially help improve the selection of individuals for Alzheimer’s disease drug trials, and the machine learning approach can be extended to model other spinal fluid-based biomarkers.
The researchers will present their work on a blood test for another key Alzheimer’s biomarker, tau, in Lisbon during a conference from 26 to 31 March.