Abstract
Early identification of individuals at risk of developing Alzheimer's disease (AD) dementia is important for developing disease-modifying therapies. In this study, given multimodal AD markers and clinical diagnosis of an individual from one or more timepoints, we seek to predict the clinical diagnosis, cognition and ventricular volume of the individual for every month (indefinitely) into the future. We proposed and applied a minimal recurrent neural network (minimalRNN) model to data from The Alzheimer's Disease Prediction Of Longitudinal Evolution (TADPOLE) challenge, comprising longitudinal data of 1677 participants (Marinescu et al., 2018) from the Alzheimer's Disease Neuroimaging Initiative (ADNI). We compared the performance of the minimalRNN model and four baseline algorithms up to 6 years into the future. Most previous work on predicting AD progression ignore the issue of missing data, which is a prevalent issue in longitudinal data. Here, we explored three different strategies to handle missing data. Two of the strategies treated the missing data as a “preprocessing” issue, by imputing the missing data using the previous timepoint (“forward filling”) or linear interpolation (“linear filling). The third strategy utilized the minimalRNN model itself to fill in the missing data both during training and testing (“model filling”). Our analyses suggest that the minimalRNN with “model filling” compared favorably with baseline algorithms, including support vector machine/regression, linear state space (LSS) model, and long short-term memory (LSTM) model. Importantly, although the training procedure utilized longitudinal data, we found that the trained minimalRNN model exhibited similar performance, when using only 1 input timepoint or 4 input timepoints, suggesting that our approach might work well with just cross-sectional data. An earlier version of our approach was ranked 5th (out of 53 entries) in the TADPOLE challenge in 2019. The current approach is ranked 2nd out of 63 entries as of June 3rd, 2020.
Highlights
Alzheimer’s disease (AD) dementia is a devastating neurodegenerative disease with a long prodromal phase and no available cure
We only showed minimalRNN with model filling (RNN–MF), linear state space (LSS) with model filling (LSS–MF), long short-term memory (LSTM) with model filling (LSTM-MF) and SVM/SVR using one input timepoint because they yielded the best results within their model classes
recurrent neural networks (RNNs)-MF was statistically better than LSTMMF for clinical diagnosis and ventricular volume, but not ADAS-Cog13
Summary
Alzheimer’s disease (AD) dementia is a devastating neurodegenerative disease with a long prodromal phase and no available cure. There is significant interest in predicting the longitudinal disease progression of individuals. Since AD dementia is marked by beta-amyloid- and tau-mediated injuries, followed by brain atrophy and cognitive decline (Jack et al, 2010, 2013), a multimodal approach might be more effective than a single modality approach to disentangle this heterogeneity and predict longitudinal disease progression (Marinescu et al, 2018, 2020). We proposed a machine learning algorithm to predict multimodal AD markers (e.g., ventricular volume, cognitive scores, etc.) and clinical diagnosis of individual participants for every month up to six years into the future. Our goal is the longitudinal prediction of clinical diagnosis and multimodal AD markers at a potentially unlimited number of timepoints
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
