Researchers at Case Western Reserve University have demonstrated a machine-learning algorithm that combines neuroimaging with neurophysiological, proteomic and genomic diagnostics to predict Alzheimer’s disease early on in its advance.
In a study published online Aug. 15 in Scientific Reports, Anant Madabhushi, PhD, and colleagues describe their work testing the algorithm with data on 149 patients in the Alzheimer’s Disease Neuroimaging Initiative.
The team is calling the algorithm Cascaded Multi-view Canonical Correlation (CaMCCo) and reporting that it incorporates “all diagnostic categories and optimizes classification by selectively combining a subset of modalities at each level of the cascade.”
Their results suggest that “fusion of select modalities for each classification task outperforms fusion of all modalities and individual modalities,” according to the study abstract. “In addition, CaMCCo outperforms all other multi-class classification methods” for predicting mild cognitive impairment.
In a news item published by the university, Madabhushi says many previous studies have compared healthy subjects with Alzheimer’s patients, “but there’s a continuum. We deliberately included mild cognitive impairment, which can be a precursor to Alzheimer’s, but not always.”