Multimodal prediction of conversion to Alzheimer's disease based on incomplete biomarkers∗This work was supported by the Bernstein Computational Program of the German Federal Ministry of Education and Research (01GQ1001C, 01GQ0851, GRK 1589/1), the European Regional Development Fund of the European Union (10153458 and 10153460), and Philips Research.∗

BackgroundThis study investigates the prediction of mild cognitive impairment-to-Alzheimer's disease (MCI-to-AD) conversion based on extensive multimodal data with varying degrees of missing values.MethodsBased on Alzheimer's Disease Neuroimaging Initiative data from MCI-patients including all available modalities, we predicted the conversion to AD within 3 years. Different ways of replacing missing data in combination with different classification algorithms are compared. The performance was evaluated on features prioritized by experts and automatically selected features.ResultsThe conversion to AD could be predicted with a maximal accuracy of 73% using support vector machines and features chosen by experts. Among data modalities, neuropsychological, magnetic resonance imaging, and positron emission tomography data were most informative. The best single feature was the functional activities questionnaire.ConclusionExtensive multimodal and incomplete data can be adequately handled by a combination of missing data substitution, feature selection, and classification.