Alzheimer's disease (AD) is known as one of the major causes dementia and characterized by slow progression over several years, with no treatments or available medicines. In this regard, there have been efforts to identify risk developing AD in its earliest time. While many previous works considered cross-sectional analysis, more recent studies focused on diagnosis prognosis longitudinal time series data a way modeling. Under same problem settings, work, we propose novel computational...

Electronic health records (EHR) consist of longitudinal clinical observations portrayed with sparsity, irregularity, and high dimensionality, which become major obstacles in drawing reliable downstream outcomes. Although there exist great numbers imputation methods to tackle these issues, most them ignore correlated features, temporal dynamics, entirely set aside the uncertainty. Since missing value estimates involve risk being inaccurate, it is appropriate for method handle less certain...

Transfer learning has attracted considerable attention in medical image analysis because of the limited number annotated 3-D datasets available for training data-driven deep models real world. We propose Medical Transformer, a novel transfer framework that effectively volumetric images as sequence 2-D slices. To improve high-level representation 3-D-form empowering spatial relations, we use multiview approach leverages information from three planes volume, while providing parameter-efficient...

In this paper, we propose a novel architecture of deep neural network for EEG-based motor imagery classification. Unlike the existing networks in literature, proposed allows us to analyze learned weights from neurophysiological perspective, thus providing an insight into underlying patterns inherent induced EEG signals. order validate effectiveness method, conducted experiments on BCI Competition IV-IIa dataset by comparing with competing methods terms Cohen's k value. For qualitative...

Abstract Major depressive disorder (MDD) is a leading cause of disability; its symptoms interfere with social, occupational, interpersonal, and academic functioning. However, the diagnosis MDD still made by phenomenological approach. The advent neuroimaging techniques allowed numerous studies to use resting‐state functional magnetic resonance imaging (rs‐fMRI) estimate connectivity for brain‐disease identification. Recently, attempts have been investigate effective (EC) that represents...

Transfer learning has gained attention in medical image analysis due to limited annotated 3D datasets for training data-driven deep models the real world. Existing 3D-based methods have transferred pre-trained downstream tasks, which achieved promising results with only a small number of samples. However, they demand massive amount parameters train model imaging. In this work, we propose novel transfer framework, called Medical Transformer, that effectively volumetric images form sequence 2D...

Electronic health record (EHR) data are sparse and irregular as they recorded at time intervals, different clinical variables measured each observation point. In this work, to handle multivariate time-series data, we consider the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">human knowledge</i> of aspects be measure them in situations, known multi-view features, which indirectly represented data. We propose a scheme realize features...

Deep neural networks perform well in artificially- balanced datasets, but real-world data often has a long-tailed distribution. Recent studies have focused on developing unbiased classifiers to improve tail class performance. Despite the efforts learn fine classifier, we cannot guarantee solid performance if representations are of poor quality. However, learning high-quality setting is difficult because features classes easily overfit training dataset. In this work, propose mutual framework...

Electronic health records (EHR) have become an important source of a patient data but characterized by variety missing values. Using the variational inference Bayesian framework, autoencoder (VAE), deep generative model, has been shown to be efficient and accurate capture latent structure complex high-dimensional data. Recently, it used for imputation. In this paper, we propose general framework that incorporates effective imputation using VAE multivariate time series prediction. We utilize...

Electronic health records (EHRs) are characterized as nonstationary, heterogeneous, noisy, and sparse data; therefore, it is challenging to learn the regularities or patterns inherent within them. In particular, sparseness caused mostly by many missing values has attracted attention of researchers who have attempted find a better use all available samples for determining solution primary target task through defining secondary imputation problem. Methodologically, existing methods, either...

Electronic health records (EHR) are characterized as non-stationary, heterogeneous, noisy, and sparse data; therefore, it is challenging to learn the regularities or patterns inherent within them. In particular, sparseness caused mostly by many missing values has attracted attention of researchers, who have attempted find a better use all available samples for determining solution primary target task through defining secondary imputation problem. Methodologically, existing methods, either...

Alzheimer's disease (AD) is known as one of the major causes dementia and characterized by slow progression over several years, with no treatments or available medicines. In this regard, there have been efforts to identify risk developing AD in its earliest time. While many previous works considered cross-sectional analysis, more recent studies focused on diagnosis prognosis longitudinal time series data a way modeling (DPM). Under same problem settings, work, we propose novel computational...

Electronic health records (EHR) consist of longitudinal clinical observations portrayed with sparsity, irregularity, and high-dimensionality, which become major obstacles in drawing reliable downstream outcomes. Although there exist great numbers imputation methods to tackle these issues, most them ignore correlated features, temporal dynamics entirely set aside the uncertainty. Since missing value estimates involve risk being inaccurate, it is appropriate for method handle less certain...

Electronic health record (EHR) data is sparse and irregular as it recorded at time intervals, different clinical variables are measured each observation point. In this work, we propose a multi-view features integration learning from multivariate series by self-attention mechanism in an imputation-free manner. Specifically, devise novel multi-integration attention module (MIAM) to extract complex information inherent data. particular, explicitly learn the relationships among observed values,...